Accessibility settings

This paper is in the following e-collection/theme issue:

Machine Learning (2997) Mobile Health (mhealth) (4980) Clinical Informatics (2106) Artificial Intelligence (4462) Precision Medicine (416)

Published on in Vol 14 (2026)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/83092, first published .
Woman checking fitness tracker during workout with kettlebell nearby

Computational Insights Into Smart Bioelectronics in Digital Health Care (2020-2024): Topic Modeling Study

Computational Insights Into Smart Bioelectronics in Digital Health Care (2020-2024): Topic Modeling Study

Authors of this article:

JiWon Bae1 Author Orcid Image ;   JiHoon Lee1 Author Orcid Image ;   Pildong Hwang1 Author Orcid Image ;   Ji Eun Shin1, 2 Author Orcid Image ;   Sung Ryul Shim1, 2 Author Orcid Image ;   Jong-Yeup Kim1, 2, 3 Author Orcid Image ;   Seunghee Lee1 Author Orcid Image
Article Authors Cited by Tweetations (1) Metrics

    1Konyang Medical Data Research Group-KYMERA, Konyang University Hospital, 158 Gwanjeodong-ro, Seo-gu, Daejeon, Republic of Korea

    2Department of Biomedical Informatics, College of Medicine, Konyang University, Daejeon, Republic of Korea

    3Department of Otorhinolaryngology-Head and Neck Surgery, College of Medicine, Konyang University Hospital, Daejeon, Republic of Korea

    Corresponding Author:

    Seunghee Lee, PhD


    Background: Smart bioelectronics are electronic medical devices that combine hardware and artificial intelligence (AI)–based software. These convergent medical devices analyze bio-signals measured through hardware using AI algorithms and deliver physical stimulation to enhance therapeutic effects.

    Objective: This study aimed to systematically analyze recent research trends in smart bioelectronics to understand their evolving role in digital health care and to provide evidence-based insights for shaping future research and development strategies.

    Methods: A total of 92 publications indexed in PubMed between 2020 and 2024 were analyzed. Latent Dirichlet allocation–based topic modeling, optimized using coherence scores, was applied to identify latent research themes.

    Results: The results indicate a steady increase in related research over the past 5 years, along with a clear shift in research focus from bio-signal sensing and bioelectronic device materials toward AI-driven analysis and disease-oriented applications, ultimately evolving into intelligent and adaptive bioelectronic therapeutic systems. Three major research topics were identified: bio-signal–based neuromodulation (n=23, 25%), AI-driven neurological disease analysis (n=32, 34.7%), and implantable bioelectronics and biomaterials (n=37, 40.2%).

    Conclusions: By mapping the evolving landscape of smart bioelectronics, this study provides valuable insights into their multidisciplinary development and highlights their potential applications in clinical decision support, personalized rehabilitation, and next-generation medical device innovation.

    JMIR Med Inform 2026;14:e83092

    doi:10.2196/83092

    Keywords

    smart bioelectronics; keyword analysis; topic modeling; research trends; PubMed 


    Background

    Bioelectronics are medical devices that provide therapeutic effects through physical stimulation without using drugs and are receiving attention in the medical community because they can solve the side effects and misuse problems associated with existing drug treatments [1]. Although existing drug treatments use chemical components to cause biochemical reactions in the body to relieve symptoms, bioelectronics directly stimulate nerves to induce therapeutic effects; therefore, their mechanism of action is clear, and side effects are relatively low. In addition, real-time treatment monitoring and immediate feedback are possible, allowing for more precise treatment.

    In the past, bioelectronics were used in a limited manner by inserting them into the body, such as pacemakers, but recently, as they have evolved into noninvasive methods, an environment is being created where patients can receive treatment at home [2]. In addition, the market is rapidly expanding as clinical efficacy has been proven for various diseases such as depression, insomnia, headache, dementia, and epilepsy. In the global market, commercialization of bioelectronics is actively progressing, centered around the United States and Europe, and the number of related products is gradually increasing. In contrast, the domestic bioelectronics market is still in its initial stages, and product development and clinical application are not active, so competitiveness is low. Accordingly, research and new technology development for effective clinical application of bioelectronics are essential in Korea as well, and it is necessary to increase treatment precision and strengthen the competitiveness of the medical device industry through the combination with artificial intelligence (AI), big data, and wearable technology.

    In the health care field, the amount of data, such as research papers, patents, and clinical data, is increasing exponentially. These data provide important research insights, but the volume is so large that manual analysis is limited. However, data analysis is essential for suggesting research directions and exploring innovative treatments. Accordingly, research trend analysis using text mining and topic modeling is actively being conducted.

    Text mining is a technology that extracts meaningful information by formalizing unstructured data and is used to quantitatively understand research trends by analyzing research papers and patent documents. Topic modeling is a technique that automatically derives key topics from large numbers of documents and is useful for classifying topics in research papers and analyzing research trends [3]. Through these approaches, it is possible to analyze which topics are currently being studied most actively and how the proportion of specific studies has changed over time. A study that analyzed the main topics of health literacy–related research identified research trends over the past 10 years and suggested future research directions [4]. In another study, the topic modeling techniques were applied to derive key technologies and major issues in the analysis of domestic research trends related to the Fourth Industrial Revolution [5]. Using these techniques can contribute to a more systematic understanding of the latest trends in electronic pharmaceutical research and help establish future research directions.

    This study aims to analyze the trends in smart electronic drug research over the past 5 years, to analyze how research on the topic of “smart electronic drugs” is progressing, and to suggest future research directions. Relevant studies were identified from the PubMed database using bioelectronics-related keywords combined with AI-related terms in the titles or abstracts. Frequency analysis and latent Dirichlet allocation (LDA)–based topic modeling were applied. First, we analyzed the quantitative research flow and meaning by identifying the trends and changes in the core keywords of smart electronic drugs research over time through frequency analysis. Next, we analyzed the main topics and distribution of the research content using LDA-based topic modeling to explore overall research trends and their implications (Figure 1).

    ‎
    Figure 1. Analytical framework for smart bioelectronics research trend analysis: the workflow began with systematic literature curation using a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)–based screening process, followed by natural language processing (NLP)–based text preprocessing, including normalization, tokenization, and stop-word removal. Latent Dirichlet allocation (LDA) was applied for topic modeling, with model evaluation and topic number optimization performed using coherence metrics. The resulting topics were further analyzed to identify temporal research trends, domain-specific classifications, and emerging artificial intelligence (AI) integration patterns, providing clinically relevant and translational insights in digital health care. DBS: deep brain stimulation: VNS: vagus nerve stimulation.

    In addition, this study set 3 research topics to identify trends in smart bioelectronics research and suggest future directions, and the following implications were derived from the analysis results of each topic:

    • Research topic 1: how have research trends in smart bioelectronics changed over the past 5 years, and what do they mean?
    • Research topic 2: what are the core keywords and main topics of smart bioelectronics research trends, and what do they mean?
    • Research topic 3: what are the characteristics of the main topics derived from the topic analysis, and what strategic directions can be derived from each topic?

    Smart Bioelectronics

    Smart bioelectronics are advanced technologies that combine AI-based software with hardware-based bio-stimulation devices to measure bio-signals in real time, analyze them with AI algorithms, and provide physical stimulation optimized for an individual’s condition, thereby maximizing therapeutic effects [6]. Unlike conventional chemical drug treatments, smart bioelectronics directly stimulate nerves and tissues to induce physiological responses, enabling precise and personalized treatment. Consequently, they are attracting attention as a new alternative for chronic diseases, neurological diseases, and rehabilitation treatment.

    Meanwhile, the term referring to electronic drugs first appeared in earnest in 2013 when GlaxoSmithKline introduced a therapeutic device that controls neural circuits through electrical stimulation and used the expression “electroceutical.” However, the terminology in this field has not been unified to date, and the terms “electroceutical” and “bioelectronic medicine” are used interchangeably, and “bioelectronics” and “electronic drugs” are also used in various ways depending on the literature. As this confusion of terms can obscure technical and academic identities, it is necessary to establish clear concepts and a classification system in future research and industrialization stages.

    Topic Modeling

    Topic modeling is a technique that uses machine learning and natural language processing technology to analyze large amounts of text data and automatically extract major topics from documents [3]. It is used to understand the content of documents, group related topics, and explore hidden patterns within data. Topic modeling is particularly useful for extracting meaningful information from massive document datasets, such as research papers, news articles, and electronic medical records, and is widely applied across various fields, including medicine and biology.

    Recently, topic modeling has been increasingly applied in the medical field to analyze large-scale biomedical data and uncover hidden patterns that provide new insights. For example, previous studies have applied topic modeling techniques to analyze public responses to the COVID-19 vaccine using social media data and identified key discussion themes such as vaccine side effects, vaccine hesitancy, and infectious disease management policies [7]. In addition, advanced topic modeling approaches such as the graph attention-embedded topic model have been applied to electronic health record data to identify potential disease-related topics and improve the interpretability of large-scale clinical datasets [8].

    Topic modeling is also widely used to identify research trends in scientific literature. For example, previous studies have applied LDA-based topic modeling to analyze research trends in precision medicine and medical device usability by examining large collections of journal papers and media data, enabling the discovery of emerging research themes and key industry elements [9,10]. In addition, studies by Özyurt et al [11-14] demonstrated effective approaches for reporting LDA parameters, selecting optimal topic numbers based on coherence scores, and interpreting temporal research trends. These studies provide important methodological benchmarks for conducting bibliometric topic modeling analysis.

    Following these approaches, this study applies LDA-based topic modeling to analyze research trends in smart bioelectronics literature.


    Data Collection

    This study used the PubMed database to collect publications related to smart bioelectronics. PubMed is a widely used biomedical database that provides access to references and abstracts in the life sciences and biomedical fields, primarily including MEDLINE-indexed literature.

    The literature search targeted studies related to bioelectronics and electroceutical research combined with AI. Because there is currently no dedicated Medical Subject Headings term for electroceutical or bioelectronic medicine, the search was conducted using keywords appearing in the titles and abstracts of papers.

    To capture the heterogeneous terminology used in this emerging research area, 2 complementary PubMed database search strategies were used. The first query focused on electroceutical and neuromodulation-related terminology combined with AI-related keywords, while the second query targeted bioelectronics together with AI, sensor, wearable, and closed-loop control concepts.

    The criteria for paper selection were as follows:

    • Studies containing bioelectronics-related keywords (eg, “bioelectronics,” “electroceutical,” or “neuromodulation”) and AI-related terms (eg, “artificial intelligence,” “machine learning,” “deep learning,” or “reinforcement learning”) in the title or abstract
    • Papers published between January 1, 2020, and December 31, 2024
    • Studies involving human subjects
    • Papers providing free full text and abstracts
    • Exclusion of review papers

    The PubMed search was conducted using the following queries:

    • Query 1: (electroceutical*[Tiab] OR neuromodulation[Tiab] OR “bioelectronic medicine”[Tiab] OR “vagus nerve stimulation”[Tiab] OR “nerve stimulation”[Tiab] OR “deep brain stimulation”[Tiab]) AND (“artificial intelligence”[Tiab] OR “machine learning”[Tiab] OR “deep learning”[Tiab] OR “reinforcement learning”[Tiab]) AND (“2020/01/01”[pdat] : “2024/12/31”[pdat]) AND humans[MeSH Terms] AND fha[Filter] NOT review[pt]
    • Query 2: (bioelectronic*[Tiab] OR electroceutical*[Tiab] OR “bioelectronic medicine”[Tiab]) AND (“Artificial Intelligence”[Tiab] OR AI[Tiab] OR “Machine Learning”[Tiab] OR “Deep Learning”[Tiab] OR sensor*[Tiab] OR wearable*[Tiab] OR “closed-loop”[Tiab] OR “closed loop”[Tiab]) AND (“2020/01/01”[pdat] : “2024/12/31”[pdat]) AND humans[MeSH Terms] AND fha[Filter] NOT review[pt]

    After applying the inclusion criteria and conducting manual screening based on titles and abstracts, 49 studies from query 1 and 43 studies from query 2 were retained. Duplicate records were checked using title and DOI matching; however, no duplicate studies were identified between the 2 query results. Consequently, a total of 92 studies were included in the final dataset for the subsequent text mining and LDA-based topic modeling analysis, as listed in Multimedia Appendix 1 [15-106].

    Ethical Considerations

    This study used publicly available bibliographic data retrieved from PubMed. No individual participant data or identifiable personal information was included. Therefore, institutional review board approval and informed consent were not required for this study.

    Data Preprocessing

    Preprocessing was performed to ensure the quality of the bibliographic data from the 92 selected papers. The preprocessing steps are as follows. First, the entire text was converted to lowercase to ensure consistency in word notation. To select only the words essential for semantic analysis, 7571 nouns were selected through headword extraction and part-of-speech tagging. Subsequently, special characters and stop words were removed, as they were unnecessary for analysis. Stop-word removal was performed using the Natural Language Toolkit (NLTK) default stop-word dictionary [12] to remove common stop words. Additionally, 2886 stop words were removed, including custom stop words such as “system,” “study,” and “patient,” which frequently appeared in papers but did not contribute to semantic differentiation. Finally, 49 single-letter words were filtered out, resulting in 4452 text data items suitable for analysis (Table 1).

    Table 1. Data preprocessing steps and main tasks.
    Preprocessing stageMain tasks
    Convert to lowercaseEnsuring text consistency
    Noun and headword extractionPerform lemmatization and extract nouns
    Remove special charactersRemove unnecessary symbols such as !, ", #, $, %, &, ’, (), and *
    Remove stop wordsNatural Language Toolkit stop words and custom stop words (eg, “system” and “study”)
    Word filteringRemove 1-letter words

    The list of custom stop words was categorized as follows. A total of 122 words with low contribution to semantic distinction were selected and classified into types as follows: (1) research and analysis, (2) technology or system, (3) AI and data, (4) medical or health, (5) sensors and devices, and (6) other terms (Table 2).

    Table 2. Custom stop-word list.
    CategoryStop words
    Research and analysisstudy, analysis, method, approach, result, performance, accuracy, effect, outcome, score, rate, parameter, process, objective, finding, design, test, validation, conclusion, research, optimization, correlation, scale, mapping, precision, AUC (Area Under the Curve), classifier, classification, regression, crossvalidation, solution, response, beta, reference, application
    Technology and systemssystem, technology, technique, development, network, tool, array, setting, delivery, management
    Artificial intelligence and datadata, model, machine, artificial intelligence, AI (Artificial Intelligence), ML (Machine Learning), SVM (Support Vector Machine), GPT (Generative Pre-trained Transformer), RL (Reinforcement Learning), layer, input, decision, baseline, ratio, difference, set, information, feature, support, forest
    Medical and healthpatient, participant, subject, cohort, treatment, trial, intervention, condition, risk, brain, motor, activity, efficacy, medication, healthcare, drug, sensation
    Sensors and devicessensor, device, biosensor, bioelectronics, stimulation, DBS (deep brain stimulation), VNS (vagus nerve stimulation), tcVNS (Transcutaneous cervical vagus nerve stimulation), tVNS (Transcutaneous Vagus Nerve Stimulation), STN (Subthalamic Nucleus), neuromodulation
    Otherpotential, group, area, case, ability, side, number, quality, level, state, time, change, addition, challenge, task, target, background, individual, rating, selection, use, value, field, function, Hz (Hertz), paper, region

    Word Clouds

    Word cloud analysis is a representative technique for visualizing text-based unstructured data. It allows intuitive identification of key keywords by varying the font size based on word frequency. In this study, frequency analysis was performed on the collected text data to identify the frequency of key keywords. The preprocessed text was tokenized into words, and keywords were extracted based on their frequency of appearance. Word frequency was calculated using the “explode()” and “value_counts()” functions in the pandas library. To ensure inclusion of all words, the minimum frequency (min_word_count) was set to 5 for extraction. On the basis of these results, visualization was performed using Python’s WordCloud library, allowing an at-a-glance identification of key keywords related to smart bioelectronics.

    LDA Algorithm

    LDA was used for topic modeling analysis. It is an unsupervised probabilistic model that represents each document as a mixture of topics and each topic as a distribution over words. The analysis was performed using the LDA algorithm implemented in the Gensim library. The model parameters were set to “iterations=100” and “random_state=4190,” and the number of topics was explored in the range of 3 to 10 with an interval of 1.

    The optimal number of topics was determined based on coherence scores, which measure the semantic consistency of keywords within each topic. Perplexity was also examined as a supplementary metric for model evaluation; however, coherence was used as the primary criterion for selecting the optimal number of topics. Furthermore, the results of the LDA models were visualized using pyLDAvis, and the main keywords for each topic were identified based on the top 10 keywords generated by the model.


    Frequency Analysis Results by Research Period

    The literature search and dataset construction process is illustrated in Figure 2. Because terminology in electroceutical and bioelectronic medicine research remains heterogeneous, some relevant studies may not have been captured by the predefined queries. Nevertheless, the 2 complementary search strategies were designed to improve coverage of this rapidly evolving research field. The annual publication trend for smart bioelectronics-related studies from 2020 to 2024 is shown in Figure 3. The number of publications increased steadily from 12 in 2020 to a peak of 23 in 2022, reflecting the growing research interest in smart bioelectronics (Figure 3).

    ‎
    Figure 2. Literature search and dataset construction process. AI: artificial intelligence.

    Using the term frequency–inverse document frequency (TF-IDF) technique, we visualized the most representative keywords for each year from 2020 to 2024 using word clouds. The results reveal a gradual evolution in research focus within the smart bioelectronics field.

    In the early stage (2020), keywords such as “biomarkers,” “electrode,” “subcallosal cingulate,” “hydrogel,” “skin,” and “ECG (Electrocardiography)” were dominant, indicating that early research primarily focused on bio-signal sensing technologies and bioelectronic device materials. These keywords reflect the importance of physiological monitoring and electrode-based interfaces in early smart bioelectronics studies. In 2021, disease-oriented terms such as “parkinson disease,” “pain,” and “disease” became more prominent, together with signal acquisition and processing technologies, including “EEG,” “algorithm,” and “control.” This pattern suggests an increasing focus on neurological disorders and the application of computational methods in bioelectronic systems. In 2022, keywords such as “prediction,” “control,” “monitoring,” “therapy,” and “signal” appeared more frequently. These terms indicate the growing integration of computational analysis, signal processing, and control mechanisms in smart bioelectronic systems, reflecting a transition toward data-driven medical device technologies. In 2023, keywords including “algorithm,” “biomarkers,” “sleep,” “tremor,” and “disease” emerged, suggesting increased interest in AI-assisted clinical monitoring and biomarker analysis for neurological and physiological conditions. Finally, in 2024, AI-related keywords such as “learning” and “prediction” became more prominent alongside clinical and device-related terms such as “surgery,” “monitoring,” “tissue,” and “electrode.” These findings indicate a growing integration of AI and predictive analytics in smart bioelectronics research, suggesting a shift toward more intelligent and adaptive bioelectronic therapeutic systems.

    ‎
    Figure 3. Number of papers by years during the study period (2020-2024). Among the study years, 2022 recorded the highest number of publications, with a total of 23 papers.

    Word Cloud Analysis Results

    Word clouds were generated from the top 10 keywords based on the bag-of-words (BoW) and TF-IDF methods. The BoW-based word cloud highlights keywords such as “Parkinson disease,” “biomarkers,” “disease,” “electrode,” “learning,” “prediction,” “algorithm,” “therapy,” “control,” and “EEG.” In contrast, the TF-IDF-based word cloud includes “Parkinson disease,” “biomarkers,” “control,” “prediction,” “electrode,” “EEG,” “disease,” “learning,” “therapy,” and “tissue.”

    These keywords can be categorized as follows based on a semantic approach. First, “Sensor/Hardware” refers to terms associated with neural interfaces and bioelectronic sensing components that enable signal acquisition from biological systems, including keywords such as “electrode” and “tissue.” Second, “Application” relates to clinical and therapeutic contexts where bioelectronic technologies are applied for disease management, including terms such as “parkinson disease” and “therapy.” Third, “Analysis/Processing” refers to computational approaches used to analyze and interpret bio-signals, including keywords such as “prediction,” “learning,” “algorithm,” and “control.” Finally, “Signal Monitoring and Biomarkers” refers to the measurement and identification of physiological indicators used to monitor neural activity and disease states, including terms such as “EEG” and “biomarkers.”

    “Parkinson disease” emerged as the most prominent keyword in both BoW and TF-IDF analyses. Accordingly, co-occurrence analysis was conducted to examine the relationships between keywords appearing alongside “Parkinson disease.” Network visualization was performed using the top 30 co-occurring keywords.

    The results revealed that the research landscape can be broadly categorized into 5 major domains: neural signal acquisition, neuroimaging, clinical symptoms, neuromodulation therapy, and AI-based biomarker analysis. These findings suggest that Parkinson disease research extends beyond clinical observation, encompassing neural signal processing, advanced imaging techniques, therapeutic interventions, and data-driven analytical approaches.

    Specifically, neural signal acquisition includes electrocardiogram, electrode, microelectrode, band, and recording; neuroimaging includes magnetic resonance imaging, functional magnetic resonance imaging, imaging, cortex, nucleus, and connectivity; clinical symptoms include tremor, gait, symptom, fluency, and impairment; neuromodulation therapy includes subthalamic nucleus, surgery, and levodopa; and AI-based biomarker analysis includes prediction, algorithm, learning, and biomarkers (Figure 4)

    ‎
    Figure 4. Co-occurrence network for keywords associated with “parkinson disease.” The network was generated based on the top 30 co-occurring keywords. AI: artificial intelligence.

    Topic Modeling Analysis Results

    The coherence score was used to evaluate the LDA model across different numbers of topics, and the optimal number of topics was determined based on the highest coherence value (Figure 5). Accordingly, 3 topics were selected for the final model, and the intertopic distance map is presented in Figure 6.

    Three main thematic areas were identified in smart bioelectronics (Table 3).

    ‎
    Figure 5. Coherence scores for different numbers of topics.
    ‎
    Figure 6. Intertopic distance map visualized using multidimensional scaling. The equations for estimated term frequency within the selected topic were obtained from Chuang et al [107] and Sievert and Shirley [108]. ALFF: amplitude of low-frequency fluctuation; ECG: electrocardiogram; fMRI: functional magnetic resonance imaging; MDD: major depressive disorder; STN: subthalamic nucleus.
    Table 3. Topic modeling results with representative keywords for each topic.
    TopicRepresentative keywords
    Topic 1: bio-signal–based neuromodulation (publications: 23/92, 25.0%)
    1. pain
    2. biomarkers
    3. nerve
    4. therapy
    5. signal
    6. tissue
    7. monitoring
    8. control
    9. skin
    Topic 2: artificial intelligence–driven neurological disease analysis (publications: 32/92, 34.7%)
    1. parkinson disease
    2. disease
    3. learning
    4. EEGa
    5. algorithm
    6. control
    7. STNb
    8. subcallosal cingulate
    9. image
    Topic 3: implantable bioelectronics and biomaterials (publications: 37/92, 40.2%)
    1. electrode
    2. prediction
    3. memory
    4. infection
    5. learning
    6. parkinson disease
    7. sleep
    8. tissue
    9. Hydrogel

    aEEG: electrocardiogram.

    bSTN: subthalamic nucleus.

    Topic 1 represents bio-signal–based neuromodulation, characterized by keywords such as “pain,” “biomarkers,” “nerve,” and “signal,” indicating a focus on physiological signal acquisition and biomarker-driven therapeutic approaches. From a technical perspective, this topic primarily relies on signal sensing and feature extraction techniques, which form the foundation for subsequent data-driven analysis. Clinically, these approaches enable continuous physiological monitoring and provide objective biomarkers for disease assessment and treatment evaluation.

    Topic 2 corresponds to AI-driven neurological disease analysis, including “Parkinson disease,” “EEG,” “learning,” and “algorithm,” suggesting the increasing use of machine learning techniques for the analysis and interpretation of neurological disorders. These approaches are commonly associated with advanced AI methodologies, such as deep learning architectures for bio-signal processing and pattern recognition. From a clinical perspective, these developments contribute to improved diagnostic accuracy, early detection of neurological conditions, and more personalized treatment strategies.

    Topic 3 reflects implantable bioelectronics and biomaterials, with representative keywords such as “electrode,” “hydrogel,” “tissue,” and “wound,” highlighting advancements in implantable devices and their interaction with biological systems. The inclusion of terms such as “prediction” and “learning” further indicates the integration of data-driven approaches with implantable systems, enabling intelligent and adaptive therapeutic functions. Clinically, these technologies support real-time monitoring, closed-loop stimulation, and improved treatment efficiency through dynamic adjustment of therapeutic parameters.

    Overall, these topics illustrate the evolving research landscape of smart bioelectronics, progressing from bio-signal–based monitoring to AI-driven analysis and, more recently, to intelligent, adaptive, and clinically applicable bioelectronic systems. This progression reflects a paradigm shift toward precision medicine, in which data-driven insights and responsive therapeutic devices are increasingly integrated to address key clinical challenges.

    The temporal distribution of topics reveals a clear evolution in smart bioelectronics research (Figure 7). Early studies were dominated by topic 3, reflecting a strong focus on bioelectronic devices and materials. Between 2021 and 2022, topic 2 became predominant, indicating a rapid expansion of AI-driven analysis and disease-oriented research. In 2023, the distribution became more balanced across topics, suggesting an integration phase [105]. By 2024, topic 3 regained prominence, highlighting a shift toward the practical application and clinical translation of intelligent bioelectronic systems.

    ‎
    Figure 7. Temporal distribution of latent Dirichlet allocation (LDA)–derived topic proportions from 2020 to 2024.

    Table S1 in Multimedia Appendix 1 summarizes the results of a preliminary review of 92 key papers related to smart bioelectronics [15-106]. These papers were published in leading journals such as Nat Commun, Adv Sci (Weinh), Sensors (Basel), and Brain Stimul. The most frequently appearing keywords were “machine learning,” “bioelectronics,” “deep brain stimulation (DBS),” and “Parkinson’s disease,” indicating that smart bioelectronics research is evolving beyond general AI-based analysis toward clinically oriented applications, particularly in neuromodulation and neurological disorder management. Although traditional machine learning models, such as support vector machines and random forests, remain the most frequently used approaches, deep learning models, including convolutional neural networks, are increasingly being adopted, indicating a gradual shift toward more advanced analytical techniques.


    Principal Findings

    This study conducted keyword analysis and topic modeling based on academic papers on smart bioelectronics published over the past 5 years. The analysis identified three major thematic areas: (1) bio-signal–based neuromodulation, (2) AI-driven neurological disease analysis, and (3) implantable bioelectronics and biomaterials. These findings demonstrate that smart bioelectronics are expanding across diverse domains, both technologically and clinically. In particular, they are emerging as a convergent technology that enables integrated disease management, spanning early detection, diagnosis, treatment, and continuous monitoring.

    Furthermore, the coexistence and intermingling of terms such as “bioelectronic medicine” and “electronic drug” highlight the need for clearer conceptual definitions and a standardized classification system in this field. Establishing consistent terminology and conceptual frameworks will be essential to improve the efficiency, interoperability, and advancement of interdisciplinary research and technological applications.

    From a technological perspective, conventional bioelectronic devices have traditionally relied on hardware-based electrical stimulation to induce therapeutic effects. In contrast, smart bioelectronics are increasingly developing into integrated systems that combine bio-signal sensing, wireless communication, and data-driven analytical capabilities. Consistent with these findings, the identified topics indicate a transition toward systems that combine continuous bio-signal monitoring, machine learning-based data analysis, and adaptive therapeutic functions.

    These technological advancements are closely linked to their clinical relevance. Bio-signal–based approaches enable continuous and objective physiological monitoring, while AI-driven analytical methods support more accurate diagnosis and interpretation of complex neurological conditions. In addition, implantable bioelectronic systems facilitate real-time monitoring and closed-loop therapeutic interventions, allowing treatments to be dynamically adjusted based on patient-specific conditions. Collectively, these developments highlight the potential of smart bioelectronics to improve diagnostic accuracy, enhance treatment personalization, and support more efficient long-term disease management.

    Despite these contributions, this study has several limitations. The dataset consisted of 92 papers, representing a moderate sample size for exploratory topic modeling. However, the relatively limited number of documents may affect the stability of probabilistic models such as LDA; therefore, the identified topics should be interpreted as indicative trends rather than definitive thematic structures. In addition, data collection was limited to the PubMed database, which may restrict the comprehensive coverage of the research landscape. Future research should expand the scope of literature collection to include multiple academic databases, such as IEEE Xplore and Embase, to improve coverage and robustness. The application of advanced natural language processing techniques, such as BERTopic, may further enhance analytical precision and enable more fine-grained topic extraction. These efforts will contribute to a more comprehensive understanding of the technological evolution and core research directions in smart bioelectronics. Smart bioelectronics represent a multidisciplinary research field encompassing medical technology, information technology, and life sciences. Continued integration of clinical research and technological innovation, along with collaboration among academia, industry, and health care systems, will be essential to accelerate the development of AI-driven medical data analysis systems and expand clinical applications, ultimately advancing next-generation precision health care.

    Conclusions

    This study analyzed recent research trends in the field of smart bioelectronics by conducting keyword analysis and LDA-based topic modeling on 92 papers collected from PubMed. The findings reveal an increasing trend in research activity over the past 5 years, with topics evolving from bio-signal sensing and device-level technologies to AI-driven analysis, predictive modeling, and clinically oriented, adaptive bioelectronic systems for precision medicine. These results provide a foundation for shaping domestic research and development strategies in smart bioelectronics and offer insights for applications in digital health care, AI-driven therapeutics, and next-generation wearable medical devices. Despite these contributions, this study is limited by the relatively small number of papers analyzed and its reliance on a single database. Future research should integrate data from multiple international databases and adopt more advanced computational approaches, such as transformer-based topic modeling and network analysis, to better capture diverse research trends and uncover emerging subfields in bioelectronics.

    Funding

    This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare, Republic of Korea (grant RS-2024-00439193).

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    Preliminary review of the paper.

    DOCX File, 33 KB
    1. Safety, efficacy, and even convenience? The criteria for electronic medicine. HiT NEWS. 2021. URL: https://www.hitnews.co.kr/news/articleView.html?idxno=36684 [Accessed 2026-05-19]
    2. A tingling sensation when attached… The “era of electronic medicine” for treating depression at home. Chosun Media. Oct 29, 2024. URL: https://www.chosun.com/economy/science/2024/10/29/HOEXUU6I4NBZ3HDUK33M4B6CPQ/ [Accessed 2026-05-19]
    3. 04. Understanding text mining analysis. Wikidocs. URL: https://wikidocs.net/254900 [Accessed 2026-05-19]
    4. Kim SH, Kang WJ, Yoon E, Lee J. Determining global trends in health literacy research using topic modeling. kjhep. Jun 30, 2023;40(2):79-91. [CrossRef]
    5. Ji-young K, Dong-jo N. Analysis of domestic research trends in the 4th industrial revolution field using topic modeling. J Korean Biblia Soc. 2023;34(4):207-234. URL: https:/​/www.​kci.go.kr/​kciportal/​ci/​sereArticleSearch/​ciSereArtiView.​kci?sereArticleSearchBean.​artiId=ART003031316 [Accessed 2026-05-28]
    6. Ministry of Health and Welfare (2024). 2024 health and medical technology research and development project request for proposal (RFP). Ministry of Health and Welfare; 2024:12. URL: https://www.mohw.go.kr/board.es?mid=a10501010200&bid=0003&act=view&list_no=1480990&tag=&nPage=1 [Accessed 2026-05-28]
    7. Roh GH, Lee S, Kim JY, Lee S. SNSMiner_VAC: analyzing vaccination based on social network service data for safety surveillance. Expert Syst Appl. Dec 2024;255(B):124684. [CrossRef]
    8. Zou Y, Pesaranghader A, Verma A, Buckeridge D, Li Y. Modeling electronic health record data using an end-to-end knowledge-graph-informed topic model. arXiv. Preprint posted online on Jun 3, 2022. [CrossRef]
    9. Ahn MK. Research trend analysis of medical devices usability studies using topic modeling. SDC. Apr 30, 2020;81(2):79-90. [CrossRef]
    10. Ahn SH, Ko JH, Lee SJ. Exploring the precision medicine research trend using topic modeling. APJCRI. Jan 31, 2024;10(1):387-398. [CrossRef]
    11. Ozyurt O, Ayaz A. Twenty-five years of education and information technologies: insights from a topic modeling based bibliometric analysis. Educ Inf Technol (Dordr). 2022;27(8):11025-11054. [CrossRef] [Medline]
    12. Ozyurt O, Ozyurt H. 30-year trends in research on enriching education and training with virtual reality: an innovative study based on machine learning approach. Educ Inf Technol. 2023;29(7):8221-8249. [CrossRef]
    13. Ozyurt H, Ozyurt O, Mishra D. Exploring the evolution of educational serious games research: a topic modeling perspective. IEEE Access. 2024;12:81827-81841. [CrossRef]
    14. Özköse H, Ozyurt O, Ayaz A. Management information systems research: a topic modeling based bibliometric analysis. J Comput Inf Syst. 2023;63(5):1166-1182. [CrossRef]
    15. Brown EC, Clark DL, Forkert ND, Molnar CP, Kiss ZHT, Ramasubbu R. Metabolic activity in subcallosal cingulate predicts response to deep brain stimulation for depression. Neuropsychopharmacology. Sep 2020;45(10):1681-1688. [CrossRef] [Medline]
    16. Castaño-Candamil S, Piroth T, Reinacher P, Sajonz B, Coenen VA, Tangermann M. Identifying controllable cortical neural markers with machine learning for adaptive deep brain stimulation in Parkinson’s disease. Neuroimage Clin. 2020;28:102376. [CrossRef] [Medline]
    17. Sendi MS, Kanta V, Inman CS, et al. Amygdala stimulation leads to functional network connectivity state transitions in the hippocampus. Annu Int Conf IEEE Eng Med Biol Soc. Jul 2020;2020:3625-3628. [CrossRef] [Medline]
    18. Gurel NZ, Wittbrodt MT, Jung H, et al. Automatic detection of target engagement in transcutaneous cervical vagal nerve stimulation for traumatic stress triggers. IEEE J Biomed Health Inform. Jul 2020;24(7):1917-1925. [CrossRef] [Medline]
    19. Martineau T, He S, Vaidyanathan R, Brown P, Tan H. Optimizing time-frequency feature extraction and channel selection through gradient backpropagation to improve action decoding based on subthalamic local field potentials. Annu Int Conf IEEE Eng Med Biol Soc. Jul 2020;2020:3023-3026. [CrossRef] [Medline]
    20. Ershad F, Thukral A, Yue J, et al. Ultra-conformal drawn-on-skin electronics for multifunctional motion artifact-free sensing and point-of-care treatment. Nat Commun. Jul 30, 2020;11(1):3823. [CrossRef] [Medline]
    21. Gurel NZ, Huang M, Wittbrodt MT, et al. Quantifying acute physiological biomarkers of transcutaneous cervical vagal nerve stimulation in the context of psychological stress. Brain Stimul. 2020;13(1):47-59. [CrossRef] [Medline]
    22. Kwon YT, Kim YS, Kwon S, et al. All-printed nanomembrane wireless bioelectronics using a biocompatible solderable graphene for multimodal human-machine interfaces. Nat Commun. Jul 10, 2020;11(1):3450. [CrossRef] [Medline]
    23. Gazi AH, Gurel NZ, Richardson KL, et al. Digital cardiovascular biomarker responses to transcutaneous cervical vagus nerve stimulation: state-space modeling, prediction, and simulation. JMIR Mhealth Uhealth. Sep 22, 2020;8(9):e20488. [CrossRef] [Medline]
    24. Lee Y, Yim SG, Lee GW, et al. Self-adherent biodegradable gelatin-based hydrogel electrodes for electrocardiography monitoring. Sensors (Basel). 2020;20(20):5737. [CrossRef] [Medline]
    25. Mahmood M, Kwon S, Berkmen GK, et al. Soft nanomembrane sensors and flexible hybrid bioelectronics for wireless quantification of blepharospasm. IEEE Trans Biomed Eng. Nov 2020;67(11):3094-3100. [CrossRef] [Medline]
    26. Rwei AY, Lu W, Wu C, et al. A wireless, skin-interfaced biosensor for cerebral hemodynamic monitoring in pediatric care. Proc Natl Acad Sci U S A. Dec 15, 2020;117(50):31674-31684. [CrossRef] [Medline]
    27. Boutet A, Madhavan R, Elias GJ, et al. Predicting optimal deep brain stimulation parameters for Parkinson’s disease using functional MRI and machine learning. Nat Commun. May 24, 2021;12(1):3043. [CrossRef] [Medline]
    28. Liebrand LC, Zhutovsky P, Tolmeijer EK, et al. Deep brain stimulation response in obsessive-compulsive disorder is associated with preoperative nucleus accumbens volume. Neuroimage Clin. 2021;30:102640. [CrossRef] [Medline]
    29. Geraedts VJ, Koch M, Kuiper R, et al. Preoperative electroencephalography-based machine learning predicts cognitive deterioration after subthalamic deep brain stimulation. Mov Disord. Oct 2021;36(10):2324-2334. [CrossRef] [Medline]
    30. Sendi MS, Inman CS, Bijanki KR, et al. Identifying the neurophysiological effects of memory-enhancing amygdala stimulation using interpretable machine learning. Brain Stimul. 2021;14(6):1511-1519. [CrossRef] [Medline]
    31. Sendi MS, Waters AC, Tiruvadi V, et al. Intraoperative neural signals predict rapid antidepressant effects of deep brain stimulation. Transl Psychiatry. Nov 3, 2021;11(1):551. [CrossRef] [Medline]
    32. Lei IM, Jiang C, Lei CL, et al. 3D printed biomimetic cochleae and machine learning co-modelling provides clinical informatics for cochlear implant patients. Nat Commun. Oct 29, 2021;12(1):6260. [CrossRef] [Medline]
    33. Geraedts VJ, Koch M, Contarino MF, et al. Machine learning for automated EEG-based biomarkers of cognitive impairment during deep brain stimulation screening in patients with Parkinson’s disease. Clin Neurophysiol. May 2021;132(5):1041-1048. [CrossRef] [Medline]
    34. Aqajari SA, Cao R, Kasaeyan Naeini E, et al. Pain assessment tool with electrodermal activity for postoperative patients: method validation study. JMIR Mhealth Uhealth. May 5, 2021;9(5):e25258. [CrossRef] [Medline]
    35. Park KH, Sun S, Lim YH, et al. Clinical outcome prediction from analysis of microelectrode recordings using deep learning in subthalamic deep brain stimulation for Parkinson`s disease. PLoS One. 2021;16(1):e0244133. [CrossRef] [Medline]
    36. Chen W, Wang M, Wang N, Du C, Ma X, Li Q. The impacts of subthalamic nucleus-deep brain stimulation (STN-DBS) on the neuropsychiatric function of patients with Parkinson’s disease using image features of magnetic resonance imaging under the artificial intelligence algorithms. Contrast Media Mol Imaging. 2021;2021:9915206. [CrossRef] [Medline]
    37. Hussain I, Young S, Kim CH, Benjamin HC, Park SJ. Quantifying physiological biomarkers of a microwave brain stimulation device. Sensors (Basel). Mar 8, 2021;21(5):1896. [CrossRef] [Medline]
    38. Fang Y, Zou Y, Xu J, et al. Ambulatory cardiovascular monitoring via a machine-learning-assisted textile triboelectric sensor. Adv Mater. Oct 2021;33(41):e2104178. [CrossRef] [Medline]
    39. Missey F, Botzanowski B, Migliaccio L, Acerbo E, Głowacki ED, Williamson A. Organic electrolytic photocapacitors for stimulation of the mouse somatosensory cortex. J Neural Eng. Nov 19, 2021;18(6). [CrossRef] [Medline]
    40. Hosseini Jafari B, Zlobina K, Marquez G, et al. A feedback control architecture for bioelectronic devices with applications to wound healing. J R Soc Interface. Dec 2021;18(185):20210497. [CrossRef] [Medline]
    41. Ruiz-Valdepeñas Montiel V, Sempionatto JR, Vargas E, et al. Decentralized vitamin C & D dual biosensor chip: toward personalized immune system support. Biosens Bioelectron. Dec 15, 2021;194:113590. [CrossRef] [Medline]
    42. Lee JM, Choi EJ, Chung JH, et al. A DNA-derived phage nose using machine learning and artificial neural processing for diagnosing lung cancer. Biosens Bioelectron. Dec 15, 2021;194:113567. [CrossRef] [Medline]
    43. Krause KJ, Phibbs F, Davis T, Fabbri D. Predicting motor responsiveness to deep brain stimulation with machine learning. AMIA Annu Symp Proc. 2022;2021:651-659. [Medline]
    44. Tekriwal A, Baker S, Christensen E, et al. Quantifying neuro-motor correlations during awake deep brain stimulation surgery using markerless tracking. Sci Rep. Oct 27, 2022;12(1):18120. [CrossRef] [Medline]
    45. Wang MB, Boring MJ, Ward MJ, Richardson RM, Ghuman AS. Deep brain stimulation for Parkinson’s disease induces spontaneous cortical hypersynchrony in extended motor and cognitive networks. Cereb Cortex. Oct 8, 2022;32(20):4480-4491. [CrossRef] [Medline]
    46. Chen Y, Zhu G, Liu D, et al. Seed-based connectivity prediction of initial outcome of subthalamic nuclei deep brain stimulation. Neurotherapeutics. Mar 2022;19(2):608-615. [CrossRef]
    47. Obst MA, Al-Zubaidi A, Heldmann M, et al. Five weeks of intermittent transcutaneous vagus nerve stimulation shape neural networks: a machine learning approach. Brain Imaging Behav. Jun 2022;16(3):1217-1233. [CrossRef] [Medline]
    48. Alhourani A, Wylie SA, Summers JE, et al. Developing predictor models of postoperative verbal fluency after deep brain stimulation using preoperative neuropsychological assessment. Neurosurgery. Aug 1, 2022;91(2):256-262. [CrossRef] [Medline]
    49. Habets JG, Herff C, Fasano AA, et al. Multicenter validation of individual preoperative motor outcome prediction for deep brain stimulation in Parkinson’s disease. Stereotact Funct Neurosurg. 2022;100(2):121-129. [CrossRef] [Medline]
    50. Chen Y, Zhu G, Liu Y, et al. Predict initial subthalamic nucleus stimulation outcome in Parkinson’s disease with brain morphology. CNS Neurosci Ther. May 2022;28(5):667-676. [CrossRef] [Medline]
    51. Hussain SJ, Quentin R. Decoding personalized motor cortical excitability states from human electroencephalography. Sci Rep. Apr 15, 2022;12(1):6323. [CrossRef] [Medline]
    52. Zhang J, Zhou C, Xiao X, et al. Magnetic resonance imaging image analysis of the therapeutic effect and neuroprotective effect of deep brain stimulation in Parkinson’s disease based on a deep learning algorithm. Int J Numer Method Biomed Eng. Nov 2022;38(11):e3642. [CrossRef] [Medline]
    53. Ma J, Wang Z, Cheng T, et al. A prediction model integrating synchronization biomarkers and clinical features to identify responders to vagus nerve stimulation among pediatric patients with drug-resistant epilepsy. CNS Neurosci Ther. Nov 2022;28(11):1838-1848. [CrossRef] [Medline]
    54. Khawaldeh S, Tinkhauser G, Torrecillos F, et al. Balance between competing spectral states in subthalamic nucleus is linked to motor impairment in Parkinson’s disease. Brain. Mar 29, 2022;145(1):237-250. [CrossRef] [Medline]
    55. Lopes EM, Rego R, Rito M, Chamadoira C, Dias D, Cunha JPS. Estimation of ANT-DBS electrodes on target positioning based on a new PerceptTM PC LFP signal analysis. Sensors (Basel). Sep 1, 2022;22(17):6601. [CrossRef] [Medline]
    56. Choi W, Heo D, Kim T, et al. Stress dissipation encoded silk fibroin electrode for the athlete-beneficial silk bioelectronics. Adv Sci (Weinh). Mar 2022;9(8):e2105420. [CrossRef] [Medline]
    57. Olçomendy L, Cassany L, Pirog A, et al. Towards the integration of an islet-based biosensor in closed-loop therapies for patients with type 1 diabetes. Front Endocrinol (Lausanne). 2022;13:795225. [CrossRef] [Medline]
    58. Choi YS, Jeong H, Yin RT, et al. A transient, closed-loop network of wireless, body-integrated devices for autonomous electrotherapy. Science. May 27, 2022;376(6596):1006-1012. [CrossRef] [Medline]
    59. Macchia E, Kovács-Vajna ZM, Loconsole D, et al. A handheld intelligent single-molecule binary bioelectronic system for fast and reliable immunometric point-of-care testing. Sci Adv. Jul 8, 2022;8(27):eabo0881. [CrossRef] [Medline]
    60. Ilfeld BM, Said ET, Abdullah B, Finneran Iv JJ. Treating intractable postamputation pain with noninvasive, wearable, nonthermal, pulsed shortwave (radiofrequency) therapy: a 12-patient case series. Am J Case Rep. Aug 17, 2022;23:e937549. [CrossRef] [Medline]
    61. Pazienza A, Monte D. Introducing the monitoring equipment mask environment. Sensors (Basel). Aug 24, 2022;22(17):6365. [CrossRef] [Medline]
    62. Patel S, Ershad F, Lee J, et al. Drawn-on-skin sensors from fully biocompatible inks toward high-quality electrophysiology. Small. Sep 2022;18(36):e2107099. [CrossRef] [Medline]
    63. Kim M, Hwang JC, Min S, et al. Multimodal characterization of cardiac organoids using integrations of pressure-sensitive transistor arrays with three-dimensional liquid metal electrodes. Nano Lett. Oct 12, 2022;22(19):7892-7901. [CrossRef] [Medline]
    64. Valle G, Aiello G, Ciotti F, et al. Multifaceted understanding of human nerve implants to design optimized electrodes for bioelectronics. Biomaterials. Dec 2022;291:121874. [CrossRef] [Medline]
    65. Peres I, Rolo P, Ferreira JA, et al. Multiscale sensing of bone-implant loosening for multifunctional smart bone implants: using capacitive technologies for precision controllability. Sensors (Basel). Mar 25, 2022;22(7):2531. [CrossRef] [Medline]
    66. Alagapan S, Choi KS, Heisig S, et al. Cingulate dynamics track depression recovery with deep brain stimulation. Nature. Oct 2023;622(7981):130-138. [CrossRef] [Medline]
    67. Borda L, Gozzi N, Preatoni G, Valle G, Raspopovic S. Automated calibration of somatosensory stimulation using reinforcement learning. J Neuroeng Rehabil. Sep 26, 2023;20(1):131. [CrossRef] [Medline]
    68. Chen H, Wang Y, Ji T, Jiang Y, Zhou XH. Brain functional connectivity-based prediction of vagus nerve stimulation efficacy in pediatric pharmacoresistant epilepsy. CNS Neurosci Ther. Nov 2023;29(11):3259-3268. [CrossRef] [Medline]
    69. Kahana MJ, Ezzyat Y, Wanda PA, et al. Biomarker-guided neuromodulation aids memory in traumatic brain injury. Brain Stimul. 2023;16(4):1086-1093. [CrossRef] [Medline]
    70. Wang DX, Ng N, Seger SE, Ekstrom AD, Kriegel JL, Lega BC. Machine learning classifiers for electrode selection in the design of closed-loop neuromodulation devices for episodic memory improvement. Cereb Cortex. Jun 20, 2023;33(13):8150-8163. [CrossRef] [Medline]
    71. Yang B, Wang X, Mo J, et al. The altered spontaneous neural activity in patients with Parkinson’s disease and its predictive value for the motor improvement of deep brain stimulation. Neuroimage Clin. 2023;38:103430. [CrossRef] [Medline]
    72. Dagnino PC, Braboszcz C, Kroupi E, et al. Stratification of responses to tDCS intervention in a healthy pediatric population based on resting-state EEG profiles. Sci Rep. May 25, 2023;13(1):8438. [CrossRef] [Medline]
    73. Sabo A, Iaboni A, Taati B, Fasano A, Gorodetsky C. Evaluating the ability of a predictive vision-based machine learning model to measure changes in gait in response to medication and DBS within individuals with Parkinson’s disease. Biomed Eng Online. Dec 11, 2023;22(1):120. [CrossRef] [Medline]
    74. Farashi S, Sarihi A, Ramezani M, Shahidi S, Mazdeh M. Parkinson’s disease tremor prediction using EEG data analysis-a preliminary and feasibility study. BMC Neurol. Nov 24, 2023;23(1):420. [CrossRef] [Medline]
    75. Takemoto A, Araki T, Nishimura K, et al. Fully transparent, ultrathin flexible organic electrochemical transistors with additive integration for bioelectronic applications. Adv Sci (Weinh). Jan 2023;10(2):e2204746. [CrossRef] [Medline]
    76. Gharibans AA, Hayes TCL, Carson DA, et al. A novel scalable electrode array and system for non‐invasively assessing gastric function using flexible electronics. Neurogastroenterology Motil. Feb 2023;35(2):e14418. [CrossRef]
    77. Park M, Yoo JY, Yang T, et al. Skin-integrated systems for power efficient, programmable thermal sensations across large body areas. Proc Natl Acad Sci U S A. Feb 7, 2023;120(6):e2217828120. [CrossRef] [Medline]
    78. Arcangeli D, Gualandi I, Mariani F, et al. Smart bandaid integrated with fully textile OECT for uric acid real-time monitoring in wound exudate. ACS Sens. Apr 28, 2023;8(4):1593-1608. [CrossRef] [Medline]
    79. Soghoyan G, Biktimirov A, Matvienko Y, Chekh I, Sintsov M, Lebedev MA. Peripheral nerve stimulation enables somatosensory feedback while suppressing phantom limb pain in transradial amputees. Brain Stimul. 2023;16(3):756-758. [CrossRef] [Medline]
    80. Ho CM, Jeong H, Lim YH, Park SJ. Effects of audio brain entrainment on Korean people with mild Insomnia. Appl Psychophysiol Biofeedback. Jun 2023;48(2):207-216. [CrossRef] [Medline]
    81. Ilfeld BM, Said ET, Gabriel RA, et al. Wearable, noninvasive, pulsed shortwave (radiofrequency) therapy for analgesia and opioid sparing following outpatient surgery: a proof-of-concept case series. Pain Pract. Jun 2023;23(5):553-558. [CrossRef] [Medline]
    82. Li Y, Yin J, Liu S, et al. Learning hand kinematics for Parkinson’s disease assessment using a multimodal sensor glove. Adv Sci (Weinh). Jul 2023;10(20):e2206982. [CrossRef] [Medline]
    83. Garg R, Driscoll N, Shankar S, et al. Wearable high-density MXene-bioelectronics for neuromuscular diagnostics, rehabilitation, and assistive technologies. Small Methods. Aug 2023;7(8):e2201318. [CrossRef] [Medline]
    84. Li H, Gong H, Wong TH, et al. Wireless, battery-free, multifunctional integrated bioelectronics for respiratory pathogens monitoring and severity evaluation. Nat Commun. 2023;14:7539. [CrossRef]
    85. de Rijk SR, Boys AJ, Roberts IV, et al. Tissue-engineered cochlear fibrosis model links complex impedance to fibrosis formation for cochlear implant patients. Adv Healthc Mater. Sep 2023;12(24):e2300732. [CrossRef] [Medline]
    86. Li Q, Lin Z, Liu R, et al. Multimodal charting of molecular and functional cell states via in situ electro-sequencing. Cell. Apr 27, 2023;186(9):2002-2017. [CrossRef] [Medline]
    87. Chang B, Geng Z, Mei J, et al. Application of multimodal deep learning and multi-instance learning fusion techniques in predicting STN-DBS outcomes for Parkinson’s disease patients. Neurotherapeutics. Oct 2024;21(6):e00471. [CrossRef] [Medline]
    88. Sun J, Sun K, Chen L, et al. A predictive study of the efficacy of transcutaneous auricular vagus nerve stimulation in the treatment of major depressive disorder: an fMRI-based machine learning analysis. Asian J Psychiatr. Aug 2024;98:104079. [CrossRef] [Medline]
    89. Alapati R, Renslo B, Jackson L, et al. Predicting therapeutic response to hypoglossal nerve stimulation using deep learning. Laryngoscope. Dec 2024;134(12):5210-5216. [CrossRef] [Medline]
    90. Pantovic A, Essert C. Evaluating the impact of reinforcement learning on automatic deep brain stimulation planning. Int J Comput Assist Radiol Surg. 2024;19(6):995-1002. [CrossRef]
    91. Li Y, Frederick RA, George D, et al. NeurostimML: a machine learning model for predicting neurostimulation-induced tissue damage. J Neural Eng. Jun 27, 2024;21(3):036054. [CrossRef] [Medline]
    92. Hussain MA, Grill WM, Pelot NA. Highly efficient modeling and optimization of neural fiber responses to electrical stimulation. Nat Commun. Aug 31, 2024;15(1):7597. [CrossRef] [Medline]
    93. Qiu J, Ajala A, Karigiannis J, et al. Deep learning and fMRI-based pipeline for optimization of deep brain stimulation during Parkinson’s disease treatment: toward rapid semi-automated stimulation optimization. IEEE J Transl Eng Health Med. 2024;12:589-599. [CrossRef] [Medline]
    94. Moser P, Reishofer G, Prückl R, et al. Real-time estimation of the optimal coil placement in transcranial magnetic stimulation using multi-task deep learning. Sci Rep. Aug 21, 2024;14(1):19361. [CrossRef] [Medline]
    95. Ciecierski KA, Mandat T. Classification of DBS microelectrode recordings using a residual neural network with attention in the temporal domain. Neural Netw. Feb 2024;170:18-31. [CrossRef] [Medline]
    96. Oliveira AL, Coelho M, Guedes LC, Cattoni MB, Carvalho H, Duarte-Batista P. Performance of ChatGPT 3.5 and 4 as a tool for patient support before and after DBS surgery for Parkinson’s disease. Neurol Sci. Dec 2024;45(12):5757-5764. [CrossRef] [Medline]
    97. Todorov D, Schnitzler A, Hirschmann J. Parkinsonian rest tremor can be distinguished from voluntary hand movements based on subthalamic and cortical activity. Clin Neurophysiol. Jan 2024;157:146-155. [CrossRef] [Medline]
    98. Gurr C, Splittgerber M, Puonti O, et al. Neuroanatomical predictors of transcranial direct current stimulation (tDCS)-induced modifications in neurocognitive task performance in typically developing individuals. J Neurosci. May 29, 2024;44(22):e1372232024. [CrossRef] [Medline]
    99. Xiong C, Dang W, Yang Q, et al. Integrated ink printing paper based self-powered electrochemical multimodal biosensing (IFP-Multi ) with ChatGPT-bioelectronic interface for personalized healthcare management. Adv Sci (Weinh). Mar 2024;11(11):e2305962. [CrossRef] [Medline]
    100. Liu Z, Song H, Lin G, et al. Wireless intelligent patch for closed‐loop in situ wound management. Adv Sci (Weinh). Aug 2024;11(29):e2400451. [CrossRef]
    101. Bolonduro OA, Chen Z, Fucetola CP, et al. An integrated optogenetic and bioelectronic platform for regulating cardiomyocyte function. Adv Sci (Weinh). Sep 2024;11(36):e2402236. [CrossRef] [Medline]
    102. Zhao X, Zhou Y, Kwak W, et al. A reconfigurable and conformal liquid sensor for ambulatory cardiac monitoring. Nat Commun. 2024;15:8492. [CrossRef]
    103. Gil B, Hall TA, Freeman DM, et al. Wireless implantable bioelectronics with a direct electron transfer lactate enzyme for detection of surgical site infection in orthopaedics. Biosens Bioelectron. Nov 1, 2024;263:116571. [CrossRef] [Medline]
    104. Liu J, Li Z, Sun M, et al. Flexible bioelectronic systems with large-scale temperature sensor arrays for monitoring and treatments of localized wound inflammation. Proc Natl Acad Sci USA. Dec 3, 2024;121(49):e2412423121. [CrossRef]
    105. Marquez G, Dechiraju H, Baniya P, et al. Delivering biochemicals with precision using bioelectronic devices enhanced with feedback control. PLoS One. 2024;19(5):e0298286. [CrossRef] [Medline]
    106. Schmidt S, Li W, Schubert M, et al. Novel high-dense microelectrode array based multimodal bioelectronic monitoring system for cardiac arrhythmia re-entry analysis. Biosens Bioelectron. May 15, 2024;252:116120. [CrossRef] [Medline]
    107. Chuang J, Manning CD, Heer J. Termite: visualization techniques for assessing textual topic models. Presented at: AVI’12: International Working Conference on Advanced Visual Interfaces; May 21-25, 2012:74-77; Capri Island, Italy. [CrossRef]
    108. Sievert C, Shirley KE. LDAvis: a method for visualizing and interpreting topics. Presented at: Workshop on Interactive Language Learning, Visualization, and Interfaces; Jun 27, 2014:63-70; Baltimore, Maryland, USA. [CrossRef]

    ‎
    AI: artificial intelligence
    BoW: bag-of-words
    LDA: latent Dirichlet allocation
    TF-IDF: term frequency–inverse document frequency

    Edited by Arriel Benis; submitted 28.Aug.2025; peer-reviewed by Mi-Hwa Song, Ozcan Ozyurt; final revised version received 30.Mar.2026; accepted 20.Apr.2026; published 23.Jun.2026.

    Copyright

    © JiWon Bae, JiHoon Lee, Pildong Hwang, Ji Een Shin, Sung Ryul Shim, Jong-Yeup Kim, Seunghee Lee. Originally published in JMIR Medical Informatics (https://medinform.jmir.org), 23.Jun.2026.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Medical Informatics, is properly cited. The complete bibliographic information, a link to the original publication on https://medinform.jmir.org/, as well as this copyright and license information must be included.