In addition to the iterative reviews conducted by the interdisciplinary team, we incorporated insights from our user study to address key aesthetic concerns (Figure 1). The main changes included removing the prominently drawn areolae, redesigning facial features to reduce an “uncanny” appearance, and adjusting contrast and color palettes to improve visual clarity. We also refined minor aspects of body proportions in response to participant feedback on overall aesthetics.

The vector illustration is structured in a clear, hierarchical way to facilitate easy manipulation and identification of body regions for pain drawing applications. At the highest level, the illustration is divided into main groups representing anatomical views—anterior (“bones-ant”) and posterior (“bones-pos”). Each view contains subgroups for distinct anatomical regions, such as “columna” (spinal column), “cranium,” “coxa” (hip region), and “superius” (upper extremities). Within each region, individual bones or vertebrae are defined as separate, clearly labeled elements within the SVG file (eg, “bones-pos-columna-C01” for cervical vertebra C01). This organized layering allows users or software developers to quickly access and highlight specific anatomical structures programmatically, greatly enhancing usability and interoperability in digital health applications.

The labeling of anatomical structures in the SVG illustration follows a systematic and standardized naming convention, ensuring clear identification and consistent referencing across applications. Each SVG element has a unique identifier directly corresponding to a specific anatomical structure. To further facilitate semantic interoperability, we mapped these identifiers to standardized clinical terminology codes from SNOMED CT. This mapping is provided in a structured JSON file, where each SVG element ID (eg, “bones-pos-clavicula-sin”) is directly associated with its corresponding SNOMED CT concept identifier (eg, 720617006). This structured mapping simplifies integration with EHRs and clinical documentation systems, enabling precise and unambiguous representation of anatomical locations in digital health and clinical informatics applications.

The final vector illustration was released under a CC BY 4.0 [26] license in the GitHub repository Open Pain Illustration [27]. Open-source availability encourages community engagement for further refinement and widespread adoption.

Quantitative data were analyzed using descriptive statistics. For each rating outcome, 2-sided 95% CIs for the proportion of participants providing a positive rating, defined as a score of 5 to 7, were calculated using the Wilson score method. Two-sided 95% CIs for means were calculated using the Student t distribution. No hypothesis tests were performed. Qualitative feedback on aesthetics and the perceived ability to indicate pain was examined through thematic analysis facilitated by the OpenAI o1-preview language model [30]. Participants’ open-ended responses were analyzed thematically to uncover key patterns, recurring themes, and actionable suggestions. Due to the large volume of textual data, the language model was used to assist in organizing and summarizing the feedback. Two separate prompts were designed for the analysis of each open-ended question (Multimedia Appendix 2), enabling the model to generate structured outputs, including recurring themes, the number of participants mentioning each theme, representative quotes, and participant identifiers for manual review. To ensure the validity of the AI-generated results, the participant identifiers were manually reviewed by one of the authors, confirming that the themes accurately reflected the responses, with no mismatches identified.

This study was conducted in accordance with ethical standards. The study synopsis was submitted to the Ethics Committee of Zurich, Switzerland, which determined that the study did not fall under the regulation of the Swiss Human Research Act (ethics ID 2022-01475). Participation was voluntary, and all participants provided electronic informed consent before beginning the survey. Survey data were collected anonymously, with no personal identifiers recorded, and securely stored.

A total of 298 participants were included in the analysis. The mean age of the sample was 46.2 (SD 15.4) years (95% CI 44.4-47.9). Of 298 participants, 154 (51.7%) were assigned female at birth, and 144 (48.3%) were assigned male at birth. Regarding back pain history, 51.7% (154/298) of the participants reported current back pain, 31.5% (94/298) had a history of back pain, and 16.8% (50/298) reported no back pain.

Detailed demographic characteristics are summarized in Table 1. In total, 0.7% (2/298) of the participants, who identified as female, were assigned male at birth. No participants who identified as male were assigned female at birth. A total of 5% (15/298) of the participants did not provide any gender description. Education and race and ethnicity distributions are also shown in Table 1.

Participants’ confidence in accurately indicating their back pain on the body chart varied across the Likert scale. Overall, 85.9% (256/298) of the participants (95% CI 81.5%-89.4%) reported a positive level of confidence (scores>4). The mean rating was 5.7 out of 7 (SD 1.2; 95% CI 5.5-5.8; N=298). In a descriptive subgroup analysis, participants with current back pain had a mean rating of 5.8 (95% CI 5.6-6.0; n=154), whereas participants in the “no back pain” category had a mean rating of 5.3 (95% CI 5.0-5.6; n=50). Participants in the “history of back pain” category (n=94) were not included in this contrast. No hypothesis test was performed. Only 5.7% (17/298) of the participants reported low confidence (scores<4), indicating that low confidence ratings were uncommon (Figure 2).

Most participants felt confident that the body chart illustration would aid communication with health care professionals. Similar to the pain indication confidence ratings, 85.6% (255/298) of the participants (95% CI 81.1%-89.1%) expressed positive confidence in this ability (scores>4), with a mean score of 5.6 out of 7 (SD 1.2; 95% CI 5.5-5.8). Only 5.7% (17/298) of the respondents reported low confidence in this ability (scores<4), as shown in Figure 2.

The illustration’s aesthetic appeal received mixed feedback: 44% (131/298) of the participants (95% CI 38.4%-49.6%) rated it positively (>4), 27.9% (83/298) rated it neutrally (4), and 28.2% (84/298) rated it negatively (<4). This variability resulted in a lower mean score of 4.3 out of 7 (SD 1.5; 95% CI 4.1-4.5), indicating room for improvement in the visual design (Figure 2).

Participants provided numerous suggestions to improve functionality and visual appeal. Although this study only presented a static illustration without any interactive features or drawing tools, many participants proposed enhancements that they would find useful in a fully functional version. Common suggestions for improving functionality included the ability to zoom or enlarge the image, add labels or anatomical landmarks, replace the skeletal view with muscular structures, and use color coding to indicate pain type or intensity. Some participants also expressed interest in additional perspectives such as side views and the ability to depict radiating or moving pain. Aesthetic feedback focused on modifying facial features, redesigning or removing nipples, adjusting body proportions, and optionally providing sex-specific versions. These insights informed final refinements and will guide future iterations. A detailed breakdown is available in Multimedia Appendix 3.

Our findings should be interpreted as results of a design validation and feasibility evaluation rather than as evidence of comparative or clinical effectiveness. Because the user study did not include an alternative body map or an objective pain marking task, it cannot determine whether the positive confidence ratings were attributable to qualities specific to this illustration or reflected more general comfort with body mapping. Within those limits, the findings suggest that an anatomically accurate, open-source vector illustration for back pain documentation was acceptable to many users and was perceived as potentially useful for identifying and communicating pain regions while also revealing areas that may benefit from further refinement. A substantial proportion of participants reported confidence in their ability to use the illustration, suggesting that its level of detail and clarity may be appropriate for many users. Individuals currently experiencing back pain reported slightly higher confidence than participants in the “no back pain” category; because this was a descriptive subgroup observation, it should be considered hypothesis generating rather than confirmatory. However, user feedback underscored the need for future iterations to address varying needs, such as additional labeling options, alternative views, and aesthetic adjustments.

Building on these insights, the qualitative feedback highlighted several opportunities to enhance both functionality and usefulness. Participants advocated for greater precision and flexibility, such as enlarging the image, enabling zoom, adding annotations, or including multiple viewpoints (eg, side views) to capture pain that radiates or shifts. These suggestions align with broader literature emphasizing usability and adaptability as key drivers of patient engagement with digital health tools [6,11,16,17]. Calls for more nuanced and realistic representations, such as incorporating muscles, also reflect the evolving understanding of how anatomical accuracy can enhance the assessment and communication of musculoskeletal pain [5].

Aesthetically, participants revealed a tension between clinical neutrality and visual appeal. While some felt that aesthetics was secondary to the tool’s medical purpose, others desired features that would humanize the figure—less unsettling facial features, optional sex-specific versions, realistic body proportions, and color enhancements. This variability underlines the importance of designing medical visualization tools that are both functionally rigorous and sensitive to users’ comfort and emotional responses, a consideration increasingly recognized in medical illustration and patient-centered design research.

Back pain is common among older adults [1-3]. Older adults and people with low digital literacy may face barriers when using digital interfaces, although the nature of these barriers may differ. Prior research indicates that digital pain drawings can be reliable [13] and that pain drawing interfaces should be tailored to users with physical or cognitive impairments [5]. In our study, the mean participant age was 46.2 (SD 15.4) years, and self-reported confidence regarding use of the illustration was generally high. However, because online recruitment required internet access and sufficient baseline digital literacy, these findings may not generalize to older adults or patients with low digital literacy. Future usability testing should therefore specifically include patients with low digital literacy, limited device access, or a need for assisted completion to assess accessibility, usability, and reliability in clinical and at-home contexts.

From a clinical informatics perspective, the integration of anatomical labels aligned with coding standards such as SNOMED CT represents a significant step toward interoperability. By ensuring that the illustrated regions can be mapped onto recognized clinical terminologies, we opened avenues for integrating patient-generated pain data into EHRs, large-scale research databases, and decision support systems. This interoperability can ultimately enrich the landscape of clinical data, supporting more consistent documentation, outcome tracking, and evidence-based decision-making in back pain management [14,15,20].

Notwithstanding these strengths, our study has limitations. Although we recruited a relatively large sample (N=298) with diverse backgrounds, online crowdsourced participants may not fully represent clinical populations or individuals with severe mobility impairments, non-English speakers, patients in specific health care contexts, or people with low digital literacy and limited internet or device access. This limitation is particularly relevant for future at-home use, where successful adoption will require interfaces, instructions, and support options that work for patients with varying digital skills. Additionally, our evaluation relied primarily on self-reported measures of confidence and perceived usefulness rather than longitudinal assessments or clinical outcomes. Future research could involve clinical trials that track how using such illustrations influences communication in consultations, diagnostic accuracy, treatment adherence, or patient satisfaction over time and should include dedicated assessments for low–digital literacy users. Furthermore, the user study included additional questions that will be explored in a future publication.

Another important limitation is that the current illustration includes only anterior and posterior views. While some participants expressed interest in having additional perspectives, particularly lateral or side views, to capture certain types of radiating or localized back pain, we have no immediate plans to add such views at this stage. Future community-driven adaptations or subsequent versions of the illustration could address this need by expanding the anatomical perspectives, but this remains outside the current scope of development. Consequently, clinicians and researchers should be aware of potential gaps in capturing pain that extends or radiates laterally.

While participants’ positive self-reported confidence suggests promise for the current illustration, it does not guarantee objective accuracy, superiority over existing body maps, or improved clinical decision-making. We acknowledge that future investigations must move beyond self-reports to assess whether patients’ markings on our illustration consistently match clinical findings or imaging results, ideally in studies with comparator body maps and predefined accuracy metrics. To that end, we are currently developing a dedicated digital tool that incorporates this body illustration and can be used on various devices (eg, personal smartphones and clinic-based tablets). This tool will allow patients to interactively mark their pain regions in real time and submit those markings securely to their health care professionals. By leveraging the SNOMED CT codes embedded within each body region, we intend to integrate these patient-drawn pain diagrams into EHR systems. Such direct linkage will enable clinicians to compare reported pain locations and severities against diagnostic imaging or clinical assessments, offering a more robust evaluation of the tool’s accuracy than subjective confidence ratings alone.

Evaluating the long-term and real-world impact of the illustration likewise remains a vital next step. Our future studies will embed the digital pain reporting tool seamlessly into clinical workflows, capitalizing on standard health data exchange protocols to transmit pain diagrams to the appropriate clinicians at the point of care. For example, patients in a pilot clinic setting may draw their pain areas at home or during triage, with these data flowing directly into the clinic’s electronic system (via Fast Healthcare Interoperability Resources– or Health Level Seven version 2–based messaging) to alert clinicians before the consultation. Such integration will allow for longitudinal tracking of how pain evolves over time, comparisons with clinical outcomes (eg, functional scores and medication use), and formal assessment of whether digital body mapping leads to better targeted interventions. By capturing these objective metrics in routine practice and pairing them with patient-reported data, we can more definitively establish whether our anatomically accurate, open-source illustration and its future associated digital application foster improvements in back pain care.

The anatomically accurate, open-source vector illustration for back pain documentation represents a feasible foundation for future digital health tools. In this design validation study, the illustration showed promise in supporting patient confidence, communication with clinicians, and integration into EHRs through alignment with coding standards such as SNOMED CT. However, the present results do not demonstrate superiority over existing body maps or objective improvements in pain localization accuracy. User feedback highlights opportunities for refinement, such as adding muscular details; optional sex-specific versions; and features such as zooming, annotations, and multiple perspectives. Moreover, future tools incorporating this illustration could further enhance usefulness by addressing functionality needs such as capturing radiating pain and offering customizable options. Iterative improvements to both the illustration and the tools that leverage it could promote broader adoption, improve standardization, and better meet diverse user needs. Future research should evaluate how the illustration and these tools impact clinical workflows and patient outcomes, ensuring their effectiveness in varied health care settings.