SOMAScience aims to address the current shortage of platforms for acquiring both deep (multidimensional and longitudinal) and wide (cross-spectrum and large-scale) pain data to create novel insights into the dynamics and evolution of acute and chronic pain. The term SOMA stems from the Greek word σόμα (meaning body or entire person), signifying that it takes a holistic approach to pain.

SOMAScience represents the combination of 3 branches: the SOMA smartphone app used for data acquisition, an application programming interface for data transfer, and open-source analysis code distributed through GitHub [39] (Figure 1).

To request to run a study through the platform, researchers need to submit a research inquiry detailing the study purpose on the SOMA website [40]. Data for SOMAScience are acquired through the SOMA app, which is freely available on Google Play [41] and the Apple App Store [42] and can be found by searching for “SOMA Pain Manager.” Anyone is able to download and use the app, regardless of whether they are participating in a research study. For associating people’s app data with a specific study, the researcher will be assigned a unique study ID (1 per study) and a list of individual app IDs (1 per expected participant), which need to be sent to the study participants. Participants can then install and register on the SOMA app and enroll in a specific study using the study ID and unique app ID (instruction videos on [43]; Figure 2). This use of study and app IDs allows the assignment of individual participants’ data to single and multicenter studies.

Upon enrollment, participants will be sent regular reminders through the app to fill out short pain surveys (details about the ESM and data content are in the following sections). At the time of publication, assessments are restricted to the features listed below. Future releases may offer the option to request additional features and questions. After each assessment, encrypted data are transferred to our application programming interface, stored on an actively managed secure database (see “Maximizing Privacy and Security” in Multimedia Appendix 1), and then shared with researchers from individual studies (Figures 1 and 2).

The APA’s app evaluation model stands as a notable benchmark for evaluating the suitability of health-related smartphone apps intended for patient populations [35]. It delineates 5 pivotal criteria to assess apps: accessibility, privacy and security, clinical foundation, engagement style, and therapeutic goal [44]. To keep pace with the rapidly advancing field of health apps, the system is regularly updated and refined [45]. Stemming from this APA model is a comprehensive database [46], which facilitates app evaluations across the 5 core domains [47]. Such initiatives are crucial in establishing public-facing, user-friendly standards for health apps and ensuring the development of safe and efficacious apps that benefit users [48-50].

In the development of SOMAScience, we have deeply integrated the cardinal principles of the APA model. Recognizing the emphasis that this model (and other akin evaluation frameworks) places on robust privacy, security, usability, and clinical foundations, we meticulously factored in specific technological elements during the app’s creation [50]. For readers interested in the technological nuances and our dedicated approaches to privacy and security aligned with the APA guidelines, we direct you to Multimedia Appendix 1 [51-60].

A multidisciplinary team at the Psychiatry, Embodiment, and Computation Lab at Brown University, comprising academic neuroscientists, psychologists, pain physicians, software engineers, and designers, collaboratively conceptualized and initialized the early design and features of the SOMA app. The SOMAScience platform is supported by the Brainstorm program at the Carney Institute for Brain Science at Brown University, a new program to accelerate the translation of computational brain science into real-world applications that benefit patients, the scientific community, and society as a whole.

Any app meant for long-term use must provide a simple, user-friendly interface tailored to its target audience. With this in mind, we actively sought initial feedback from individuals experiencing chronic pain to shape our app’s delivery structure. Through comprehensive one-on-one Zoom (Zoom Video Communications, Inc) discussions with a select patient group (n=4), representing a diverse pain spectrum including migraines, postsurgical musculoskeletal pain, cancer pain, and inflammatory bowel disease, we gleaned insights into our early app prototypes. These discussions involved a detailed walkthrough of each interface, where patients aired their thoughts in a guided interview.

Drawing from this feedback, we refined the “Check-In” interface’s design and flow. A common sentiment among initial users was the importance of daily tracking of pain and mood fluctuations in tandem with daily activities. To accommodate this, we introduced the “Trends” screen, a visual tool designed not only to foster self-management and a deeper understanding of pain dynamics but also to facilitate effective communication with health care providers. To further enhance the app’s utility, we incorporated screens to monitor various treatment modalities and transitioned the mood and emotion tracking interfaces to use intuitive visual analog scales (VASs) in lieu of a 2D rating system.

Responding to the patient’s desire for a more personalized experience, we introduced an interactive chatbot during the onboarding phase. This chatbot briefly engages users, gathering foundational demographic details and an introductory snapshot of their pain experiences. Existing studies vouch for the efficacy of chatbots in extending support to people with chronic pain [61], making this an evolving component with forthcoming features in SOMA that focus on pain interventions.

Our iterative refinement strategy incorporated a beta-testing phase. Initially, 30 internal testers actively engaged with the app daily across a gamut of devices and operating platforms, enabling us to identify and correct technical glitches and enhance the user experience, especially regarding the “Trends” data visualization. To expand our feedback, SOMA was then shared with a larger patient interest group (over 250 people with chronic pain), leading to critical refinements and the inclusion of user-suggested enhancements. This ongoing feedback mechanism ensures the continuous improvement and evolution of SOMA.

With the rising number of health-focused smartphone apps, there is also a growing need for transparency in the selection of measures for the app. Here we briefly detail the process of selection for each measure in SOMAScience as assessed through the SOMA app, its basis in the scientific and clinical literature, and what gaps it was chosen to address. The goal is to provide transparency in the design and selection process to facilitate the development of research studies using SOMAScience. Specific measures may be refined over time with user feedback and as scientific studies using SOMAScience identify areas for improvement.

Multidimensional pain assessment is a crucial aspect of clinical practice as it helps determine the effectiveness of treatment and recovery. The SOMA app assesses the pain intensity scale in addition to pain interference and pain unpleasantness to provide a more holistic measure of daily pain symptom load [87]. Participants are asked to rate each pain question on a sliding VAS, providing a simple and comparable assessment of daily pain experience (Figure 4). This operationalizes a practice advocated in the International Statistical Classification of Diseases and Related Health Problems for providers to use a composite rating of pain intensity, distress, and interference of pain to determine overall burden [88]. The composite pain score provides a multidimensional solution that balances the limitations of the unidimensional pain scale while still being easily quantifiable. The ability to analyze the 3 measures individually and as a composite score helps identify meaningful individual variability, enabling specific predictions between individuals. For example, pain unpleasantness could be more predictive of future pain in individual A, while pain interference may be more important for individual B’s long-term outcomes.

Importantly, SOMA’s 3 pain questions were chosen because they are directly comparable to results from established pain questionnaires, such as the Brief Pain Inventory [89] or the McGill Pain Inventory [90]. They also satisfy the standards set by major scientific and regulatory bodies, such as the IMMPACT recommendations, the NIH Helping to End Addiction Long-term initiative, and the FDA guidelines for assessing multidimensional components of pain [13,28,29,87]. In this way, SOMA’s multidimensional pain assessment of intensity, unpleasantness, and interference can provide important supplemental measures that are directly comparable to established clinical benchmarks and standards of care. This is critical for researchers looking to establish and validate novel pain biomarkers or end points.

Pain localization is an important aspect of pain assessment. Conventional methods of measuring pain location in medical appointments and research studies involve having individuals indicate it on a body map, such as the Brief Pain Inventory [89], the McGill Pain Questionnaire [91], or the Michigan Body Map [92]. This approach can pinpoint differences in peripheral and central pain pathology based on the localization and stability of pain representation over time. For instance, nociceptive or inflammatory pain is usually precisely localized somatically and does not change much over time, while neuropathic or chronic primary pain is often experienced in multiple bodily locations, radiates, or changes over time [93].

More recent methods of digital quantification, like the ones used on the SOMA app, have established the reliability and validity of body maps for pain assessments [94,95]. Interactive body maps delivered through digital or tablet apps are more effective than traditional paper or laptop assessments [92,96]. Yet a review of smartphone apps that use the body map for tracking pain found that few actually quantified the location ratings or provided any summary feedback [97]. The SOMA app’s interactive body map offers 46 different discrete location options on the front and back of the body that participants indicate in every daily check-in (Figure 5 and Multimedia Appendix 1). The use of discrete points ensures uniformity across devices and accounts for differences in participants’ finger size or dexterity. The “Trends” section of the SOMA app displays the body map with the percentage of times a location has been selected, enabling users to visualize the frequency of pain at a given location. For participants who experience nonspecific, difficult-to-localize, or widespread pain, such as fibromyalgia, there is an additional option to indicate “My pain is everywhere” on the body map.

The treatment of pain has been incredibly difficult to get right. The newest clinical guidelines advocate the use of multimodal, multidisciplinary approaches [27,98,99]. Such approaches emphasize a combination of pain treatments that include medications, restorative therapies (eg, physical therapy), interventional procedures (eg, epidural injections), behavioral interventions (eg, cognitive behavioral interventions), and complementary and integrative medicine (eg, acupuncture). Combinations of these therapies have been associated with the best long-term pain outcomes [100,101] and satisfy a biopsychosocial approach to pain [102].

It can be challenging for individuals and providers to determine which treatments are most effective for them, as the effects of many treatments for pain may not become apparent for weeks or even months (eg, cognitive and physical interventions, certain medications, and surgery) [103,104]. In determining how to measure treatments through the SOMA app, we followed the recommendations of the 2019 Department of Health and Human Services Pain Strategy [27].

In the case of medications, many different pathophysiologic mechanisms are targeted with different classes of medications. The use of different medications often changes over time, so we designed the medication screen on the SOMA app to be able to capture such changes. We, therefore, included 20 options across the main classes of pain medications for both acute and chronic pain on the SOMA app, detailed further in Multimedia Appendix 1.

A second treatment screen includes the recommended nonpharmacologic approaches to pain. Combinations of these treatments are often used by a single person over time to target different pain mechanisms [105]. SOMAScience currently provides the ability for people to track up to 20 different nonpharmacologic therapies across these 4 major classes, detailed in Multimedia Appendix 1. SOMAScience’s broad treatment tracking capabilities therefore facilitate the type of wide data needed to understand differences in treatment use across users.

Emotion and pain are fundamentally related. Definitions of pain acknowledge that pain is partially an emotional experience [106,107], yet few researchers would reduce pain down to just another emotional state such as sadness or happiness. The complexity of the pain-emotion relationship is highlighted by reviews of the neural circuits of each construct, showing both shared and functionally dissociable brain regions [108,109]. Unfortunately, the theoretical and empirical understanding of how emotion and pain are connected is limited, as much previous research only focuses on cross-sectional correlations between the 2 constructs [110,111].

Despite the renaissance of emotion research since the 1960s [112], emotion researchers continue to disagree on what constitutes an appropriate emotion measure [113]. The discrete emotion perspectives suggest that specific emotions such as anger, fear, happiness, sadness, disgust, and surprise are special kinds of biologically distinct responses associated with unique behavioral, physiological, and experiential correlates [114,115]. The dimensional perspectives consider emotions to be organized along a set of common dimensions such as valence (unpleasantness) and arousal (intensity) that are combined with cognitive and cultural knowledge to form emotional states.

When designing an emotion list for the SOMA app, we wanted a set of emotions that are commonly experienced, had a balance of positive and negative emotions, and were relevant to pain experiences (eg, worry). We started by selecting emotions experienced more than 5% of the time from a large-scale ecological momentary assessment study [116], removing rare emotions such as contempt or awe. Next, we balanced the remaining emotion set by using positive and negative emotions from the clinically relevant scale (Positive and Negative Affect Schedule) [117]. Lastly, we allowed participants to indicate no emotion (neutral) or emotion not listed (other).

A complementary method from a dimensional perspective is to assess general mood through a single-dimensional VAS of valence ranging from very unpleasant to very pleasant. Although there are other dimensional scales that could also be assessed, such as arousal [118,119] or goal congruence [120], valence is known to capture the majority of variance between emotion states [121], as self-reports of emotions tend to be highly correlated within a positive or negative valence [122]. Accordingly, on the SOMA app, we ask users to evaluate their current, past, or future mood on this VAS, ranging from unpleasant (0) to very pleasant (100). One major benefit of this measure is that we can quantify daily emotional experiences even if the participant does not self-report any discrete emotions from our finite list, reducing the need to interpolate or remove missing data from our analysis of the dynamics of emotion and pain (Figure 4).

People’s experiences of both pain and emotion are intrinsically tied to the activities they are engaging in each day. There have been many large-scale studies investigating the relationship between emotions and behavior, showing that physical exercise is meaningfully associated with reduced mental health difficulties [123] and that people’s choice of activities is motivated by minimizing negative affect and maximizing positive affect [124,125]. In particular, people seem to engage in mood-elevating activities (eg, socializing) when they are feeling down and mood-depressing activities (eg, work and chores) when they are feeling up. At the same time, being in pain affects both mood and daily activities.

Most people intuitively reduce their activities when they experience new-onset pain. Acute pain generally functions as an alarm bell in the brain to signal tissue damage, with the urge to rest considered a protective mechanism to prevent further injury and promote healing [126]. However, outside of the initial acute phase, a lack of activity can hinder long-term recovery and may signal underlying changes in affective and motivational brain circuits that have been causally linked to the transition from subacute to chronic low back pain [126,127]. The synergistic impact of activity engagement on mood, pain, and physical function is why activity engagement is promoted by pain self-management and rehabilitation programs alike [128]. While many people assume that patients with chronic pain move less than those not in pain, objective actigraphy data do not differ between patients with chronic pain and no-pain controls [129,130]. What has been less studied is the types and range of daily activities and how they change between acute and chronic pain stages. While people may recognize changes in their daily activities as a result of pain (eg, no longer walking as much), they may not have insight into the relationship between pain, mood, emotion, and activities (ie, which activities increase or decrease pain in the short vs long term). It remains unclear whether certain patterns of activity engagement at different stages of pain experience are important for long-term outcomes.

For this reason, we added an activities-tracking feature alongside mood and pain tracking. The activities screen on the SOMA app contains 20 activities that are known to have a dynamic relationship with mood based on large-scale, longitudinal data sets [125] or to be common among patients with pain (eg, medical visits). Specific activities selected are detailed in Multimedia Appendix 1. In addition to having the person report either momentary or daily activities, we also have participants reflect on how each activity they completed made them feel that day and how much it affected their pain. Taken together, these measurements provide a full picture of participants’ daily activities and help understand the potential bidirectional relationships between emotion, pain, and behavior. Users are also able to visualize their ratings of how much a given activity affected their pain and mood over time in the dedicated trends screen, which may help people develop insight into how certain activities help or hinder their recovery.

Expectations play an undeniably large role in pain perception. In artificial laboratory settings where healthy participants receive painful stimulation, a wealth of findings show that expectations about pain can increase the pain experience (nocebo effect) or conversely decrease it (placebo effect) [131,132]. This has important implications for the experience of pathological pain [133,134], where expectations about pain treatment are reliable predictors of treatment response [135]. These studies suggest that an ongoing cognitive modulation of pain is an important determinant of ongoing pain perception [136]. Expectations for pain relief most likely shift over time, the longer someone has experienced pain the more difficult the pain is to control [137,138]. Subtle longitudinal changes in expectations are believed to occur as pain becomes chronic, but capturing these changes in research studies is challenging. For this reason, we included an assessment of daily predictions about pain, mood, and activities on the SOMA app to capture how predictions vary over time within the same person or between different types of users (eg, patients with acute vs chronic pain).

We decided to assess expectations on the SOMA app by asking users to predict their expected levels of pain intensity, unpleasantness, interference, mood, and activities for the next day using the same scales used to capture their actual rating for that day. This allows the assessment of the bilateral influence of pain and mood expectations on actual experiences of pain, mood, and activities that are entered the following day.