{"paper_id":"aa900bba-435f-48a4-8d89-eb74e2593a1d","body_text":"ARTICLE IN PRESS\nhttps://doi.org/10.1038/s41746-026-02965-z\nReceived: 3 August 2025\nAccepted: 26 June 2026\nCite this article as: Zelovich, T.,\nZelovich, A., Feiglin, S. et al. Patient-\ntailored symptom tracking in\nendometriosis: a framework to\nexplore disease variability and\nburden. npj Digit. Med. (2026).\nhttps://doi.org/10.1038/\ns41746-026-02965-z\nTamar Zelovich, Alon Zelovich, Shaked Feiglin, Vered Klaitman-Mayer, Noémie Elhadad,\nRonya Rubinstein, Ronit Endevelt, Pinchas Akiva, Maytal Bivas-Benita & Chen Yanover\nWe are providing an unedited version of this manuscript to give early access to its\nfindings.  Before final  publication, the manuscript will undergo further editing. Please\nnote there may be errors present which affect  the content, and all legal disclaimers\napply.\nIf this paper is publishing under a Transparent Peer Review model then Peer\nReview reports will publish with the final  article.\n© The Author(s) 2026. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,\nadaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source,\nprovide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included\nin the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative\nCommons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission\ndirectly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.\nnpj Digital Medicine\nArticle in Press\nPatient-tailored symptom tracking in\nendometriosis: a framework to explore disease\nvariability and burden\n\n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nPatient-Tailored Symptom Tracking in Endometriosis:  \nA Framework to Explore Disease Variability and Burden \nTamar Zelovich1, Alon Zelovich1, Shaked Feiglin1, Vered Klaitman-Mayer2,3, Noémie Elhadad4, Ronya \nRubinstein1, Ronit Endevelt5, Pinchas Akiva1, Maytal Bivas-Benita1, Chen Yanover1  \n \n1KI Research Institute, Kfar Malal, Israel; \n2Faculty of Health Sciences, Ben Gurion University of the Negev, Beer Sheva, Israel \n3Maccabi Health Services South District, Israel \n4Columbia University Irving Medical Center, Department of Biomedical Informatics, New York, NY, \nUSA. \n5School of Public Health, University of Haifa, Haifa, Israel \n \nAbstract \nEndometriosis is a chronic, estrogen -dependent inflammatory disease characterized by highly \nindividualized, multisystemic symptoms , affecting approximately 10% of females . In this work, w e \npresent a long -term prospective study in which 34 participants defined personalized symptom sets and \nseverity scales, tracking them daily for up to 12 months using a custom -developed app. This patient-\ntailored, high -resolution monitoring revealed substantial heterogeneity in the number of symptoms \ntracked, with participants tracking a median of 24.5 unique symptoms (interquartile range [19.25, 36.5]). \nTo standardize assessment of system -organ involvement, symptoms were mapped to the MedDRA \nhierarchy, allowing structured analysis of symptom distribution. Based on these data, we propose a \nframework for characterizing symptom burden, trajectory, and disease burden, concepts that together \ncapture the variability, systemic nature, and complexity of endometriosis. This personalized approach \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \noffers a clearer understanding of the disease experience and lays the groundwork for future tools that may \nimprove communication with healthcare providers and inform more personalized and effective treatment \nstrategies. \n \n \n \n \n \n \n \n  \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nIntroduction  \nEndometriosis is a chronic, systemic, inflammatory and estrogen -dependent gynecological disease \nthat affects about 10% of women  of reproductive age worldwide .1–3 Globally, it is  estimated that 170 \nmillion women  are living with endometriosis.4 Despite its high prevalence, the average delay from \nsymptom onset to diagnosis , which often requires surgical confirmation , has been estimated  at 4 –12 \nyears.3,5–8  Endometriosis is characterized by endometrial-like cells growing outside the uterus, primarily \nin the pelvic area. These cells respond to hormonal changes, and, similar to menstruation, they may bleed, \nleading to inflammation, scar tissue formation, and adhesions that bind pelvic tissues and organs.9,10  \nThe complexity of managing endometriosis arises from its wide range of symptoms affecting multiple \norgan systems11, with high heterogeneity in symptom presentation among patients.2,12–15 Symptoms may \ninclude se vere pain during menstruation and or ovulation, pelvic pain, lower back pain, pain during \nintercourse, digestive16–20 and urinary disturbances, fatigue 12, radiating leg pain21–23, migraines16,17,24,25, \nmood swings, and infertility.26,27 This variability, along with the fluctuating intensity of symptoms \nthroughout the m enstrual cycle, makes d isease management particularly challenging, requiring highly \nindividualized treatment approaches.28–32 \nThese complexities also create significant barriers to effective patient -physician communication. 33 \nMany patients struggle to accurately present the full spectrum of their symptoms. Tracking and describing \nthese symptoms over time can be overwhelming, often leading to incomplete reporting or omission of key \ndetails. Therefore, it is often difficult to provide a comprehensive picture of a disease that fluctuates over \nmonths, leaving patients feeling unable to fully convey the breadth of their condition without fear of being \ndismissed or misunderstood. 34,35 Physicians, in turn, face additional challenges due to time constraints \nduring consultations, limiting their ability to thoroughly address every symptom reported. 36–38 Without \nsystematic tools to collect and prese nt symptoms and comorbidity data, gaining a comprehensive \nunderstanding of the disease remains difficult. This communication gap between patients and \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nphysicians39–43 can result in inadequate treatment decisions, ineffective disease management 44, and the \npotential of worsening symptoms and associated comorbidities.33,45 \nPrevious studies have explored the use of digital platforms in endometriosis, highlighting their value \nin tracking symptoms, improving patient -physician communication, and enhancing disease \nmanagement.28,34,46–52  Some platforms allowed tracking of dozens of symptoms and even enabled \ncustomized entries, with long-term usage studied across large populations over periods of six mon ths or \nmore. However, even with these technologies, symptom tracking was often based on predefined or semi-\ncustomized categories, limiting the ability to fully capture the complexity and individuality of each \npatient's experience. Studies with longer durat ions faced additional challenges, such as significant data \ngaps due to the lack of mechanisms to encourage consistent symptom tracking.28,34,48–53  \nThis study investigates how long-term, personalized symptom tracking can be used to characterize the \ncomplexity and multisystemic burden of endometriosis at the individual level. Specifically, we explored \nsymptom heterogeneity across patients, the diversity of organ systems involved, and the extent to which \nstructured symptom classification and longitudinal visu alization can capture this complexity. Beyond \ncharacterizing symptom heterogeneity, this study aims to provide a conceptual and methodological \nframework for investigating individualized symptom dynamics and disease burden in endometriosis. We \nconducted a long-term prospective study involving 34 endometriosis patients who were instructed to log \nsymptoms daily using a custom-developed app. Guided by our research team, each participant defined a \npersonalized set of symptoms and rating scales, enabling individualized longitudinal tracking over time. \nThis approach allowed us to construct a distinct “disease profile” for each individual, reflecting their \nspecific constellation of symptoms, severity, and the systems affected. We further analyzed thes e data \nusing a system -level classification framework to group symptoms by affected organ systems. This \nprovided a clearer representation of disease complexity and demonstrated how personalized app-based \ntracking can reveal the systemic burden and heterogen eity of endometriosis. While this study does not \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nassess the impact on patient –physician communication or analyze symptom patterns  over time,  it \nestablishes a critical foundation for future research in those areas. \nResults  \nThe study cohort included 34 participants. Table  Error! Reference source not found.  \nsummarizes the cohort characteristics. The median participant age was 33 years (interquartile range, IQR, \n[30.25,36.75]). The median age at first endometriosis symptoms was 14 years [12.75,17], and the median \nage at diagnosis was 29 years [24,31]. Seventeen participants (50%) had a confirmed diagnosis of \nadenomyosis, while the remaining 17 either did not have it or were unsure. Two participants (6%) reported \nknown fertility problems, six (18%) reported no impact on fertility, and 26 (76%) reported that they did \nnot know their fertility status at the time of the study. Supplementary Table S1 provides additional cohort \ncharacteristics. \nStudy Characteristics and Symptoms Variation : Table Error! Reference source not found.  \npresents statistics o f the study attributes, including the average number of reported days and hormonal \ncycles, as well as characteristics of  the tracked symptoms and their hierarchical classification within \nMedDRA.54,55 Figure 1 and Supplementary Figure S1 and Figure S2 illustrate the distribution of these \nattributes, while detailed participant-level data is provided in Supplementary Table S2. As shown in Table \nError! Reference source not found.  and Figure 1A, the median engagement rate (see Methods for \ndefinition) was 95.3% [ 76.8%,97.4%] and 74% of participants achieved an engagement rate of 80% or \nhigher. Table Error! Reference source not found. and Figure 1B-C also show that the median number \nof MedDRA’s Lowest Level Terms (LLTs) tracked by participants was 28 [20.2,40.8], affecting 10 [9,12] \nSystem Organ Classes (SOCs). In total, participants tracked 228 distinct symptoms, with each participant \nexperiencing a unique set of symptoms, highlighting the complexity and heterogeneity of endometriosis \nsymptomatology. According to the MedDRA hierarchy, these 228 participant-selected symptoms were \n\nARTICLE IN PRESS\n \n \n \nmapped to 147 LLTs, 123 Preferred Terms (PTs), 112 High-Level Terms (HLTs), 65 High-Level Group \nTerms (HLGTs), and 17 SOCs entries. \n \nFigure 1: Distribution of a selected set of study -related attributes from Table Error! Reference source \nnot found. . The y -axis represents the number of participants. Panels illustrate: (A) Participant \nengagement, defined as the ratio of days with reports for at least half of the symptoms to the total days in \nthe study; (B) and (C) Counts of distinct Lowest Level Terms (LLTs) and System Organ Classes (SOCs), \nrespectively, to which tracked symptoms are mapped. \nTable Error! Reference source not found.  presents the frequency of SOCs, derived from \ncategorizing the symptoms monitored by participants. For each SOC, the three most frequently tracked \nLLTs are reported. Because individual LLTs can be linked to more than one SOC, some symptoms, such \nas dizziness, appear in multiple categories. The number next to each LLT indicates how many participants \nmonitored that particular symptom. The most frequently reported LLTs in our cohort include headache \nand migraine, fatigue, low back pain, among others. While LLT -level data highlight specific symptoms, \nanalysis at the SOC level underscores the multisystemic nature of endometriosis, with frequent \ninvolvement of reproductive system and breast disorders, the gastrointestinal system, musculoskeletal and \nconnective tissue disorders, and other systems. A comprehensive list of symptom frequencies, both at the \napp-defined level and across all MedDRA hierarchy levels, is available in Supplementary Data 1. \nARTICLE IN PRESS\n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nConsistency of symptom value distributions across cycles and patients: We computed \npairwise Jensen –Shannon divergences (JSD) 56 between per -cycle symptom -intensity distributions \n(Methods). Figure 2 shows the resulting cycle -level matrix (cycles grouped by pat ient, ordered \nchronologically within patient) and the patient-level matrix obtained by averaging cycle-pair JSDs across \nall cycles of each patient dyad, for two representing symptoms. For the symptom feeling of heaviness , \ntwo clusters of patients show both low within-patient divergence (consistent reporting across cycles) and \nlow between-patient divergence (similar distributional profiles): patients 42 and 48, and patients 51, 59, \nand 84. Patient 61, by contrast, shows low within -patient divergence but high divergence from all other \npatients, indicating consistent self-reporting against an idiosyncratic personal baseline. For muscle pain / \nflu-like symptoms, patients 59 and 72 form a comparable within - and between-patient consistency pair. \nPatients 39 and 44 show consistent within-patient reporting on profiles distinct from the rest of the cohort.  \nOverall, within-participant JSD values were significantly lo wer than between-participant values \nin 30 of 33 symptoms (91%) reported by at least three participants (Supplementary Data 3). Symptom \ndistributions are thus more consistent within individuals than across them, motivating a participant -level \napproach to symptom-trajectory analysis. \n \n\nARTICLE IN PRESS\n \n \n \n \nFigure 2: Within- and between -patient divergence of symptom -intensity distributions across \nmenstrual cycles. Pairwise Jensen–Shannon divergence (JSD) computed between distributions of daily-\nreported intensity scores per cycle, for two representative symptoms: feeling of heaviness and muscle pain \n/ flu-like symptoms. Top row: Cycle-level JSD matrices: each cell represents the JSD between intensity \ndistributions in a pair of menstrual cycles, with cycles grouped by patient and ordered chronologically \nwithin patient; block-diagonal regions therefore correspond to within-patient cycle pairs and off-diagonal \nregions to between-patient cycle pairs. Bottom row: Patient-level JSD matrices: each off-diagonal cell is \nthe mean of square root of cycle-pair JSDs between the cycles of two patients, and each diagonal cell is \nthe mean of square root of within-patient cycle -pair JSD. Low JSD indicates similar intensity \ndistributions, high JSD indicates divergent distributions. \nPatient-Level Analysis - Representative Example (Patient ID 59): To further examine symptom \npresentation at the individual patient level, we present a representative participant from the study cohort. \nARTICLE IN PRESS\n\nARTICLE IN PRESS\n \n \n \nThis exam ple illustrates the range of symptom types monitored by the participant and shows how \nlongitudinal tracking enables detailed characterization of symptom dynamics and overall disease burden. \nPatient 59 tracked her symptoms for 256 days with a 96.9% engagement rate, covering 8 menstrual cycles. \nThe average menstrual cycle length was 31.8 days . Over this period, patient 59 monitored 26 symptoms \n(see Supplementary Data 2 for full symptoms list ). Figure 3 presents 7 of these app-tracked symptoms \n(shown at the bottom) along with their five ancestral MedDRA levels. It demonstrates how the divers e \nrange of symptoms can be classified into body systems, offering insight into the multisystemic nature of \nendometriosis. For instance, “waist pain” is categorized under both 'Gastrointestinal Signs and Symptoms' \nand 'Renal and Urinary Disorders' at the SOC level (see Supplementary Data 2 for the full set of symptoms \nfor all participants, classified according to the MedDRA hierarchy). \n \nFigure 3: Illustration of 7 out of the 26 symptoms tracked by Patient 59, categorized according to the \nMedDRA hierarchy. This visualization provides an ontological representation of the patient's symptom \nprofile, structured by meaning and clinical domain rather than time. Dots are color -coded to represent \ndifferent MedDRA hierarchy levels as follows: participant -defined symptom n ame, as entered and \ndisplayed in the app  – red, LLT – orange, PT – yellow, HLT – light green, HLGT – green, and SOC – \nblue.  \nFigure 4 presents nine representative symptoms of Patient 59 over five consecutive cycles (cycles 3–\n7), measured on a 0–4 scale. Although this patient tracked symptoms across eight full cycles, we chose to \npresent cycles 3 through 7 to avoid overwhelming the visualization and because cycle 1 was unusually \nARTICLE IN PRESS\n\nARTICLE IN PRESS\n \n \n \nlong, which would have affected the consistency and clarity of the overall presentation. Monitoring \nsymptoms throughout the entire hormonal cycle across multiple cycles, provides valuable insights into \nhow symptoms evolve over an extended period of time . For example, abdominal cramping was most \nprominent during menstruation, ovulation, and premenstrual syndrome (PMS), whereas lower back pain \nappeared more persistent throughout the cycle. Sore feet worsened in the last cycle, and fatigue is \nidentified as the most impactful symptom, as it persists throughout the entire cycle, varying in intensity. \n \nARTICLE IN PRESS\n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nFigure 4: Plots showing 9 of the 26 symptoms monitored by Patient 59 (noted above each heatmap), \nfocusing on 5 consecutive cycles out of 8 (cycles 3 –7) to highlight insights from long-term tracking. All \nsymptoms are measured on a 0–4 scale; however, for these symptoms, this patient only reported intensities \nranging from 0 to 3. Black triangles mark the last day of the cycle. White days indicates a missing report, \nwhile white days following the black triangles represent days that fall outside the actual length of that \ncycle. The numbers to the right of each cycle represent the symptom burden, defined as the percentage of \nsymptomatic days out of reported (non-missing) days for each symptom within that cycle. \nIn Figure 4, the numbers to the right of each plot indicate the percentage of symptomatic days for \neach symptom within a given cycle, defined as the proportion of reported days with severity above 0, \nproviding a quantitative measure of symptom burden  over time . Figure 5 extends this analysis by \ngrouping symptoms according to MedDRA's HLGT category to evaluate  symptom group burden; for \ninstance, within the HLGT category of Gastrointestinal signs and symptoms (as detailed in the MedDRA \nhierarchy presented in Figure 2). Patient 59 experienced 12%-33% of symptomatic days per cycle due to \n\"abdominal cramps\" and 15%-55% due to “waist pain” . However, when considering all her \nGastrointestinal signs and symptoms at the HLGT level collectively (i.e., abdominal cramps, waist pain, \nheartburn, nausea, and feeling of heaviness), the percentage of symptomatic days rises dramatically to \n75%–100%. This higher percentage represents the symptom group burden  for the HLGT category of \nGastrointestinal signs and symptoms , reflecting the persistent and widespread nature of symptoms \nthroughout the cycle. While a single symptom may appear sporadically, the combined presence of \nmultiple symptoms suggests that Patient 59 may suffer from continuous gastrointestinal discomfort nearly \nevery day of the cycle. This approach highlights the importance of a comprehensive clinical perspective: \nevaluating symptoms in isolation underestimates the true impact of the disease, whereas category -level \nanalysis reveals the full extent of symptom and disease burden on participant’s quality of life.  \n \n\nARTICLE IN PRESS\n \n \n \n \nFigure 5: Percentage of symptomatic days for Gastrointestinal signs and symptoms at the HLGT level \n(i.e., symptom group burden) along with their contributing symptoms at the app level of Patient 59 (i.e., \nsymptom burden). The black solid line represents the HLGT sy mptom group burden, reflecting the \ncombined contribution of all five descendant symptoms.  Solid, colored lines denote symptoms included \nin Figure 4, while dotted grey lines indicate symptoms not shown in Figure 4. Percentages are calculated \nusing reported (non-missing) days only.  The grey rectangle highlights cycles 3–7, which are presented in \nFigure 4 (the MedDRA hierarchy for Gastrointestinal signs at the HLGT level is fully presented in Figure \n2). \n \nARTICLE IN PRESS\n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nDiscussion \nThis study establishes the feasibility of a patient -tailored, mobile-based approach to long-term daily \nsymptom tracking in endometriosis. Using data collected from 34 individuals over periods of up to 12 \nmonths, we demonstrate how individualized symptom re porting can be structured and scaled to capture \nthe complexity and multisystemic nature of the disease. The high median number of symptoms per patient \n(24.5 [19.25, 36.5]) and the extensive number of affected SOCs (10 [9,12]), many extending far beyond \nthe pelvic region, further underscore the systemic nature of endometriosis.3,28,46,57  \nTo guide the quantification and interpretation of individualized symptom dynamics in endometriosis, \nwe propose a conceptual framework that introduces key terms : symptom trajectory, symptom burden, \nsymptom group burden,  and disease burden.  Using a patient -tailored symptom tracking approach, \nparticipants frequently reported both well-documented and under -recognized symptoms, providing \ninsights into the heterogeneous and multisystemic nature of endometriosis. Consistent with existing \nliterature, commonly reported symptoms included  lower back pain, fatigue, and dysmenorrhea. 58–62  In \naddition, participants tracked a broader range of symptoms and comorbidities than standardized \ninstruments typically capture, including  eye allergies, cold sensitivity, Sjogren ’s symptoms \n(neck/throat/eyes), genital herpes, ear pain, vocal cord problems, mouth ulcer s, and more. The \nindividualized structure of the tracking approach also allowed participants to define symptoms using their \nown terminology and distinctions. Participants emphasized that this fostered a sense of ownership and \nrelevance, which appeared, based on informal feedback, to enhance engagement. For example, they often \ndistinguished between specific pain types (e.g., left  vs. right pelvic pain, “stabbing” vs. “dull”) 63 and \nseparated similar experiences, such as pain at different stages of intercourse, even when these mapped to \nthe same MedDRA LLT. This underscores the uniqueness of each patient’s disease profile.  Importantly, \nthe inclusion of these symptoms does not establish them as endometriosis -attributable manifestations. \nGiven the clinical heterogeneity of endometriosis, we deliberately did not restrict symptom capture a \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \npriori. The breadth of symptoms documented here is consistent with prior reports of extensive comorbidity \nprofiles in endometriosis11 and highlights candidate symptoms for targeted follow -up. These findings \nwarrant further investigation in larger and more representative populations. More broadly, they highlight \nthe importance of systematically exploring the full spectrum of symptoms and the underl ying \npathophysiological mechanisms of endometriosis. A deeper understanding of these mechanisms and their \nmultisystemic manifestations could support the development of more targeted treatments and, ultimately, \nimprove patients’ quality of life. \nTo enable s tructured analysis of these highly individualized symptom reports, we integrated digital \nsymptom tracking with the MedDRA hierarchy, mapping participant-reported symptoms to standardized \nmedical terms. Its hierarchical structure enables clinical synthesis by organizing symptoms into broader \ncategories, allowing for structured analysis of systemic involvement. While symptoms were standardized \nfor aggregate analysis, the original inputs remained deeply personal and participant -driven. We view \nMedDRA standardization not as limiting personalization, but as a bridge between patient -reported data \nand clinical frameworks, supporting both individualized care and population -level insight into the \nmultisystemic nature of endometriosis.  \nEndometriosis patients face a significant challenge in effectively communicating the fluctuations and \nseverity of their symptoms throughout the hormonal cycle. To address this, we introduce two analytical \nconcepts, symptom burden and symptom trajectory, wh ich together provide a comprehensive overview \nof symptom dynamics, severity, and persistence  over time.  Consistent with the distribution -level JSD \nanalysis, demonstrating greater within -patient than between -patient similarity, the use of personalized \nsymptom scales preserves sensitivity to individual symptom dynamics and supports within -participant \nlongitudinal analysis. It does , however, limit direct comparison of absolute severity across individuals. \nWhile this study did not assess clinical outcomes directly, these concepts lay a foundation for future tools \naimed at improving disease monitoring and individualized care. \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nBy defining symptom burden in terms of symptomatic days (i.e., days with symptom scores above \nzero), we capture the full spectrum of p ain and discomfort across the cycle. Notably, what qualifies as a \nsymptomatic day is determined based on each participant’s personalized pain scale, ensuring that the \nmeasure reflects individual variations in pain sensitivity and perception .  To assess dis ease burden \ncomprehensively, we analyze both individual symptoms and aggregated symptom groups as defined by \nthe MedDRA hierarchy. This dual approach captures the broader multisystemic impact of endometriosis, \nrecognizing that improvement in a single sympt om may not equate to meaningful relief if others persist \nor worsen. Ultimately, the goal is to reduce disease burden by increasing the proportion of asymptomatic \ndays. Therefore, it is sometimes necessary to examine symptom groups as categorized by MedDRA, rather \nthan analyzing only individual symptoms, in order to estimate the cumulative impact of the disease on \noverall quality of life. \nAnalyzing symptom trajectories over extended periods supports investigation of symptom dynamics \nand variability at the in dividual level. In addition, the distribution -level JSD analysis demonstrated that \ndistributional consistency is stronger within than between individuals, a pattern that motivates patient -\nlevel analysis of symptom trajectories.  Notably, inter-patient similarity was symptom -dependent rather \nthan globally preserved across all symptoms. This  analysis does not incorporate temporal information \nwithin the menstrual cycle and therefore does not directly assess phase-specific dynamics.  \nWe conducted an in -depth single -case analysis of a representative participant, supplemented by a \ncross-participant comparison to identify shared structural characteristics.  Future research may involve \nlongitudinal analysis of individual cases to identify cyclical patterns, follo wed by comparisons across \nparticipants to detect shared or divergent trajectories. This bottom-up approach, from individual patterns \nto population -level insights, may uncover hidden symptom dynamics and support more effective, \npersonalized disease management.  \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nThis study has a few limitations.  First, throughout the study, our research team maintained close \ncommunication with participants, offering guidance on symptom and scale definitions and conducting \nmonthly check-ins. While these interactions may be perceived as interventions, potentially influencing \nsymptom reporting, they were essential for sustaining engagement and ensuring the accuracy of long-term \ndaily data collection. \nSecond, daily symptom tracking itself poses challenges. Participants somet imes skipped entries on \nasymptomatic days to avoid thinking about pain or simply due to forgetfulness, and on symptomatic days \nbecause the symptoms were too overwhelming. To minimize dropouts and ensure data completeness, we \nsent reminders when participant s had not logged symptoms for more than three consecutive days and \nallowed retrospective data entry for up to seven days. These strategies, along with regular check-ins, likely \ncontributed to the high engagement rate observed in the study.  Accordingly, the observed engagement \nreflects a supported research setting designed to enable complete longitudinal data collection and may not \ndirectly generalize to real -world use without similar support structures.  In addition, the study was not \ndesigned to determine t he minimum duration of tracking required to capture meaningful symptom \ndynamics. Multiple consecutive cycles are likely necessary; however, defining the optimal tracking \nduration remains an important direction for future research. \nThird, we did not recruit a control group to track monthly, potentially hormone-related symptoms and, \nas a result, cannot make relative claims about the symptom burden of endometriosis compared to healthy \nindividuals or those with other chronic conditions. Beyond the technical cha llenges such control groups \npresent – e.g., healthy individuals are less likely to consistently monitor symptoms, especially mild or \nnon-disruptive – this study was primarily designed to demonstrate the value of personalized symptom \ntracking, rather than to compare endometriosis to other conditions. Moreover, this cohort is small and not \nintended to be representative of the broader endometriosis population. Accordingly, our observations \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nshould not be interpreted as estimates of symptom prevalence. Future research may incorporate matched \ncontrol groups to better identify which symptom patterns are unique to endometriosis. \nFinally, life circumstances occasionally disrupted symptom tracking. Common disruptions included \nvacations, pregnancy, endometriosis surgery, and major life events such as job loss, divorce, or extended \ntravel. Moreover, the study took place in Israel between 2023 and 2025, a period marked by war and \nnational stress, which further contributed to interruptions in tracking.  Notably, no participants left the \nstudy due to initiation of hormonal treatment. \nIn conclusion, this study underscores the multifaceted nature of endometriosis and emphasizes the \nneed for individualized symptom monitoring to better understand disease profile, progression and burden. \nAcross our dataset, we observed substantial variability in symptom profiles, trajectories, and overall \ndisease burden. A detailed case study illustrates how symptom dynamics can differ across cycles, \nrevealing both persistent and ep isodic symptoms. Our personalized symptom tracking approach, \nstructured using the MedDRA hierarchy, provides a framework for capturing the complexity of \nlongitudinal symptom dynamics . By introducing new terminology  – namely, symptom trajectory, \nsymptom burden, and disease burden – we propose a conceptual framework for understanding the broader \nimpact of endometriosis at both the symptom and systemic levels.  We believe this framework can be \nextended beyond endometriosis to other complex, multisystemic or hor mone-related conditions, where \nsymptom heterogeneity and fluctuating trajectories challenge diagnosis and treatment. Personalized, \nstructured symptom tracking has the potential to uncover hidden dynamics, guide targeted interventions, \nand ultimately support more informed clinical decision-making. \nMethods \nThis prospective study involved endometriosis patients monitoring a self-defined set of symptoms and \nscales daily over a period of up to 12 months using a custom-developed mobile app.  Home ovulation test \nkits were used to determine hormonal stages for each patient, and at the end of each hormonal cycle, \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \ncumulative symptom reports were generated and shared with participants to encourage continued \nengagement and provide patients with the option to share their data with their care team. \nEthical Approval: This study was approved by the Haifa University ethical committee (Reference \nnumber 367/23). All participants provided written informed consent prior to participation in the study. \nRecruitment of Participants:  Participants were recruited through multiple channels, including \nFacebook groups of private health professionals (e.g., nutritionists, physiotherapists), Instagram accounts \nof influencers whose primary audience consists of women of reproductive age (in fie lds such as fashion, \nfood, wellness, and fitness), and in the clinics of physicians specializing in endometriosis. Participants \nwere not compensated for their participation in the study.  \nStudy Population: Participants were eligible if they had a formal diagnosis of endometriosis made \nby a recognized endometriosis specialist, in accordance with Israeli clinical guidelines 64, which do not \nrequire surgical confirmation. Specialists were identified based on the list published by the Endometriosis \nFoundation of Israel. 64 Additionally, participants were eligible if they were not using any form of \nhormonal treatment during the study period; had spontaneous (non -hormonally regulated) menstrual \ncycles with typical cycle lengths between 20 –35 days prior to enrollment; had not entered menopause; \nand, if they had undergone endometriosis surgery, at least three months had passed since the procedure.  \nStudy Design: Prior to the start of the study, interested individuals participated in an initial screening \ncall with the research team. During this call, the study objectives were explained, consent and legal \ndocumentation were reviewed, and data privacy and confidentiality mea sures were discussed. Each \nparticipant’s symptoms and overall condition were assessed to determine eligibility based on the study’s \ninclusion criteria. \nIf eligible and willing to participate, individuals first completed a medical questionnaire, followed by \na second interview to define their personalized symptom profile and select appropriate intensity scales. \nWith our guidance, participants identified the symptoms most relevant to their experience and chose how \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nto rate them, typically using 0–10 or 0–4 numerical scales, representing a range from ‘symptom -free’ to \n‘symptom at its highest intensity.46  \nFollowing the initial setup, participants used the app to track their symptoms daily over a two -week \nfamiliarization period. This allowed them to get comfortable with the app’s structure and features. At the \nend of this period, participants were given the opportunity to revise their symptom list or adjust the rating \nscales associated with each symptom. Next, they defined verbal anchors to describe what each number on \nthe scale represented in terms of intensity or impact. This process allowed for a more obje ctive and \nconsistent interpretation of symptom ratings while preserving individual nuance and relevance in daily \ntracking. Additionally, all participants tracked their general physical and emotional condition daily using \na 1–10 or 1–4 scale, with personalized verbal anchors defined in a manner similar to the pain scales, where \n1 represented the worst day and the highest value represented the best day . This personalized scaling \napproach differs from validated instruments such as the EHP -3065,66, which are administered at discrete \nintervals (e.g., every few months) to assess quality -of-life impact, or experience sampling method \n(ESM)50 approaches that capture high-frequency data over short study windows (e.g., multiple reports per \nday over several days) using predefined symptom lists. In contras t, our framework is designed for \nsustained, longitudinal daily tracking across extended periods. Participants also tracked any medications \nor treatments used to manage pain. A free -text option allowed them to note anything additional that \noccurred on a given day (further details on daily data monitoring and personalization can be found in the \nSupplementary Note 2). Once each participant finalized their symptom list and personalized scales, these \nsettings were locked for the duration of the study to ensure c onsistency in data collection ; however, \nparticipants could add new symptoms if they emerged during the study period or remove symptoms that \nwere no longer relevant. This time point was considered the official start date of the study period for that \nparticipant. From that point on, participants tracked their symptoms daily, reporting a numerical value for \neach symptom in their personalized list.  \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nTo monitor hormonal stages, participants were asked to log whether they were menstruating (yes/no), \nand to rate menstrual bleeding volume using a personalized 0–4 scale.63 Each participant defined the levels \nof bleeding based on their own experience and product usage (e.g., tampons, pads, panty liners). To track \novulation, they used a commercial ovulation kit at home, starting on day 9 and con tinuing through day \n20. If no rapid increase in luteinizing hormone (LH) levels (i.e., peak) was detected, they were instructed \nto test until one appeared, then monitor for a drop in LH levels over the following 2-3 days. This approach \naimed to enhance the accuracy of hormonal stage identification. This information was collected to provide \ntemporal reference points for each participant’s data but was not incorporated into the analyses presented \nin the current study. \nTo support sustained data collection, the  research team contacted participants who failed to submit \ndata for more than three consecutive days to check in, ensure their well -being, and provide a reminder. \nThe app allowed participants to backfill symptom data for up to seven days to accommodate occ asional \nlapses. Some participants took planned breaks (e.g., due to vacations or personal reasons), which they \ncommunicated in advance. For analysis, we concatenated each participant’s active reporting periods and \nexcluded these gaps to maintain continuity  in the data.  To quantify engagement, we classified a \nparticipant-day as “engaged” if at least 50% of tracked symptoms were logged, excluding non -symptom \nentries (e.g., medications, alternative -medicine treatments, and sick -day annotations). Participant -level \nengagement was defined as the proportion of engaged days across the participant’s total study duration. \nAt the end of each hormonal cycle, we generated an updated cumulative symptom report for each \nparticipant, summarizing all data collected from the b eginning of the study up to that point. After each \nreport was sent, we conducted a remote video interview with the participant to review the findings \ntogether, ensure the report accurately reflected their experience, and encourage continued participation. \nData Collection Through a Digital App:   To simplify the documentation process and enhance \nengagement and adherence, we developed a custom tracking app to collect patient-reported data. The app \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nincluded a calendar view designed to support user engagement and provide visual feedback. This calendar \nused distinct visual markers to indicate various event types and levels of data completeness, such as days \nwith reported menstrual bleeding; days designated for ovulation testing; days with complete data entries; \ndays with partial data entries; and days with no submitted data (further details on the digital app can be \nfound in the Supplementary Note 2). \nDatabase and Privacy:  All participants’ data  were stored in a secure, cloud -based PostgreSQL \ndatabase using a dual-schema structure to ensure privacy. Personally identifiable information was stored \nseparately from symptom data and was accessible only to authorized researchers. Each participant was \nassigned a unique identifier, which linked their anonymized symptom records for analysis. Further details \non data privacy and management are provided in the Supplementary Note 1.  \nSymptom Classification:  To enable standardized analysis, we mapped the  individualized 228 \nsymptoms and conditions  to the Medical Dictionary for Regulatory Activities (MedDRA ®)54,55, an \ninternationally recognized hierarchical medical terminology developed under the auspices of the \nInternational Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use \n(ICH; MedDRA® trademark is registered by ICH). We selec ted MedDRA due to its detailed structure, \nbroad clinical scope, and compatibility with symptom classification across multiple organ systems. Its \nhierarchical framework, from Lowest Level Terms (LLTs) to System Organ Classes (SOCs), allowed us \nto analyze symptoms across different levels of granularity and to compare our findings with other clinical \nand regulatory datasets. Mapping was manually performed by the research team, selecting the LLT closest \nto the participant-defined description. When mapping was not straightforward, decisions were discussed \nwith the clinical advisor and resolved based on the clinical context provided during onboarding. In cases \nwhere multiple LLTs were considered clinically appropriate, one-to-many mappings were retained. Each \nsymptom was then positioned within the full MedDRA hierarchy, progressing from LLT to Preferred \nTerm (PT), High-Level Term (HLT), High-Level Group Term (HLGT), and ultimately to the SOCs. For \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nexample, as shown in Figure 3, the participant chose to monitor a symptom described as ‘feeling of \nheaviness’ (i.e., abdominal heaviness). This term was linked to two LLTs: ‘feeling of fullness in abdomen’ \nand ‘feeling of gastrointestinal fullness’. Both map to the same PT: ‘abdominal distension’, under the \nHLT: ‘flatulence, bloating and distension’, HLGT: ‘gastrointestinal signs and symptoms’, and ultimately \nthe SOC: ‘gastrointestinal disorders’. The complete symptom-to-MedDRA mapping table is provided as \na Supplementary Data 2. A detailed description of the MedDRA mapping procedure is provided in the \nSupplementary Note 3. \nData Analysis: Data analysis was conducted in two stages: (i) a distribution-level similarity analysis \nto assess consistency of symptom reporting across cycles and participants, and (ii)  case study of  \nlongitudinal analyses to characterize symptom dynamics over time. \nFor the first stage, we assessed the consistency of symptom reporting across menstrual cycles by \nanalyzing the distribution of symptom severity values at the cycle level. For each symptom, we computed, \nfor each patient and each cycle, the empirical distrib ution of reported severity values (0–4). Cycles with \nmore than 10% missing daily reports were excluded, and patients with fewer than three cycles were \nremoved from the analysis. Distributional differences were quantified using the Jensen –Shannon \ndivergence (JSD), a symmetric, bounded dissimilarity measure derived from the Kullback –Leibler \ndivergence, taking values in [0, 1] under log base 2, with 0 indicating identical distributions and 1 \nindicating maximal divergence. 56 Pairwise JSD values were computed between all cycles across all \npatients, including within- and between-patient comparisons. To derive patient-level divergence, square \nroot JSD values were averaged across all pairwise cycle comparisons between two patients .67 For each \nsymptom, we compared within - and between -participant JSD distributions using a two -sided Mann -\nWhitney U test. The resulting p -values were corrected for multiple comparisons using the Benjamini –\nHochberg false discovery rate (FDR) procedur e68; symptoms with q < 0.05 were consider ed statistically \nsignificant. \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nFor the second stage, we performed within -participant longitudinal  analyses using three \ncomplementary approaches. First, we presented symptom trajectories, displaying the daily participant-\nreported appearance and severity of i ndividual symptoms across several consecutive cycles. This \napproach highlights fluctuations in symptom intensity over time and enables comparisons across cycles. \nIt helps identify specific periods that may require targeted interventions and assess changes following \ninterventions. Next, we introduced the concept of symptom burden, which quantifies the percentage of \nsymptomatic days out of reported days within each cycle , defined here, as proof of concept, as any day \non which a symptom was reported with a value above 0.  To reduce sensitivity to participant-specific scale \ndefinitions, burden was operationalized using a binary threshold (>0), making it invariant to differenc es \nin scale range (e.g., 0 –4 vs 0 –10). All metrics were computed within participants and across observed \ndays only, thereby limiting the impact of intermittent missing data and avoiding assumptions of cross -\nparticipant comparability. Accordingly, these mea sures are designed to capture within -subject temporal \ndynamics rather than absolute severity differences between individuals. While a uniform threshold was \nused here, alternative thresholds may be explored in future work to better reflect individual sympto m \nsignificance. As we aim to quantify the overall impact of the disease, it is sometimes necessary to assess \nthe cumulative burden imposed by a cluster of related symptoms. We therefore defined symptom group \nburden, for a given MedDRA level, where a day is  considered ‘symptomatic’ if at least one symptom \nwithin that group is symptomatic (here, had a value above 0 ).  As we move higher up the MedDRA \nhierarchy, we gain a broader perspective on the distribution of symptomatic versus asymptomatic days \nacross grouped symptoms. When aggregating all groups and symptoms together, we refer to the overall \ndisease burden . These metrics are intended for within -participant longitudinal analysis and are not \ndirectly comparable across individuals due to the use of personalized symptom scales and definitions. \n \nAuthor Contribution \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nT.Z. was responsible for patient recruitment, communication with participants, and conducting interviews \nthroughout the study. T.Z. and C.Y. jointly developed the methodology and study protocol, perf ormed \ndata analysis, and interpreted the results. A.Z. developed the front end and back end of the app, designed \nand implemented the database structure and functionality, and provided technical support. S.F. supported \nparticipant communication and study logistics. V.K.M. provided clinical interpretation of the results. N.E. \nsupervised the study and contributed to protocol design.  R.R. oversaw ethical compliance.  R.E. \ncontributed to ethical approvals and study  design. M.B.B. and P.A. contributed to the study structure, \nprotocol development, and overall project management.  All authors were major contributors to writing \nthe manuscript and reviewed and approved the final version. \nCompeting Interests  \nT.Z. and A.Z. are founders of SymptoLab, which is developing the application used in this study. The \nother authors do not have a competing interest. \nData Availability \nData generated or analyzed during this study are included in this published article and its supplementary \ninformation files. Additional data are available via the PhysioNet 69 platform \n(https://physionet.org/content/ki-endolist/1.0.0/), including participant -reported symptoms, a \ncomprehensive symptom-to-MedDRA mapping table, and participant-level mappings. 70 \nAcknowledgements \nNo funding was received for this research.  \nReferences \n1. Parente Barbosa, C., Bentes De Souza, A. M., Bianco, B. & Christofolini, D. M. The effect of Hormones on \nEndometriosis Development. Minerva Ginecol 63, 375–386 (2011). \n2. Wei, Y ., Liang, Y ., Lin, H., Dai, Y . & Yao, S. Autonomic nervous system and inflammation interaction in \nendometriosis-associated pain. Journal of Neuroinflammation 17, 80 (2020). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n3. Hugh S Taylor, Alexander M Kotlyar, & Valerie A Flores. Endometriosis is A Chronic Systemic Disease: Clinical \nChallenges and Novel Innovations. The Lancet 397, 839–852 (2021). \n4. Hormoz Nassiri Kigloo a et al. Endometriosis, Chronic Pain, Anxiety, and Depression: A Retrospective Study \nAmong 12 Million Women. Journal of Affective Disorders 346, 260–265 (2024). \n5. Hadfield, R., Mardon, H., Barlow, D. & Kennedy, S. Delay in the diagnosis of endometriosis: a survey of \nwomen from the USA and the UK. Hum Reprod 11, 878–880 (1996). \n6. Simoens, S. et al. The Burden of Endometriosis: Costs and Quality of Life of Women with Endometriosis and \nTreated in Referral Centres. Human Reproduction 27, 1292–1299 (2012). \n7. Surrey, E., Soliman, A. M., Trenz, H., Blauer-Peterson, C. & Sluis, A. Impact of Endometriosis Diagnostic \nDelays on Healthcare Resource Utilization and Costs. Adv Ther 37, 1087–1099 (2020). \n8. Sawsan As-Sanie et al. Endometriosis A Review. JAMA (2025). \n9. Krina T. Zondervan, D.Phil, Christian M. Becker, & Stacey A. Missmer. Endometriosis. The new england \njournal o f medicine 382, 1244–1256 (2020). \n10. Andrew W Horne, Stacey A Missmer. Pathophysiology, diagnosis, and Management of Endometriosis. BMJ \n379, 070750–070769 (2022). \n11. Tamar Zelovich, Miriam Erenberg, Vered Klaitman-Mayer, & Chen Yanover. Unveiling Endometriosis Hidden \nComorbidities Using a Data-Driven Approach: A Retrospective Matched Cohort Study. npj Womens Health 3, \n30. \n12. Sinaii, N., Cleary, S. D., Ballweg, M. L., Nieman, L. K. & Stratton, P . High rates of autoimmune and endocrine \ndisorders, fibromyalgia, chronic fatigue syndrome and atopic diseases among women with endometriosis: a \nsurvey analysis. Hum Reprod 17, 2715–2724 (2002). \n13. Yang, M.-H. et al. Women With Endometriosis Are More Likely to Suffer From Migraines: A Population-Based \nStudy. PLoS One 7, e33941 (2012). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n14. Laganà, A. S. et al. Anxiety and depression in patients with endometriosis: impact and management \nchallenges. Int J Womens Health 9, 323–330 (2017). \n15. Ramin-Wright, A. et al. Fatigue - a symptom in endometriosis. Hum Reprod 33, 1459–1465 (2018). \n16. Isabelle M. McGrath, Grant W. Montgomery, & Sally Mortlock. Insights from Mendelian randomization and \ngenetic correlation analyses into the relationship between endometriosis and its comorbidities. Human \nReproduction Update 29, 655–674 (2023). \n17. Isabelle M. McGrath, Grant W. Montgomery, & Sally Mortlock. Genomic Charactarisation of the Overlap of \nEndometriosis with 76 Comorbidities Identifies Pleiotropic and Causal Mechanisms Underlying Disease Risk. \nHuman Genetics 142, 1345–1360 (2023). \n18. Tore, U. et al. Diagnosis of Endometriosis Based on Comorbidities: A Machine Learning Approach. \nBiomedicines 11, 3015 (2023). \n19. Ek, M. et al. Gastrointestinal Symptoms Among Endometriosis Patients -A Case Cohort Study. BMC Womens \nHealth 15, 59 (2015). \n20. Yang, F. et al. Evidence of Shared Genetic Factors in the Etiology of Gastrointestinal Disorders and \nEndometriosis and Clinical Implications for Disease Management. Cell Rep Med 4, 101250 (2023). \n21. Thiel, P ., Kobylianskii, A., McGrattan, M. & Lemos, N. Entrapped by Pain: The Diagnosis and Management of \nEndometriosis Affecting Somatic Nerves. Best Pract Res Clin Obstet Gynaecol \nhttps://doi.org/10.1016/j.bpobgyn.2024.102502 (2024) doi:10.1016/j.bpobgyn.2024.102502. \n22. Allyson Augusta Shrikhande. The Consideration of Endometriosis in Women with Persistent Gastrointestinal \nSymptoms and a Novel Neuromusculoskeletal Treatment Approach. Arch Gastroenterol Res 1, 66–72 (2020). \n23. Adriana Silva de Barros et al. Musculoskeletal Evaluation of the Lower Pelvic Complex in Women with \nEndometriosis: A Case-Control Study. European Journal of Obstetrics & Gynecology and Reproductive Biology \n299, 317–321 (2024). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n24. Gretchen E. Tietjen et al. Endometriosis Is Associated With Prevalence of Comorbid Conditions in Migraine. \nHeadache 47, 1069–1078 (2007). \n25. Antonio Sarria-Santamera, Yerden Yemenkhan, Milan Terzic, Miguel A. Ortega, & Angel Asunsolo del Barco. \nA Novel Classification of Endometriosis Based on Clusters of Comorbidities. Biomedicines 11, 2448 (2023). \n26. Bonavina, G. & Taylor, H. S. Endometriosis-associated infertility: From pathophysiology to tailored \ntreatment. Front Endocrinol (Lausanne) 13, 1020827 (2022). \n27. Bhurke, A. V. et al. Clinical characteristics and surgical management of endometriosis-associated infertility: A \nmulticenter prospective cohort study. Int J Gynaecol Obstet 159, 86–96 (2022). \n28. Ensari, I., Pichon, A., Lipsky-Gorman, S., Bakken, S. & Elhadad, N. Augmenting the Clinical Data Sources for \nEnigmatic Diseases: A Cross-Sectional Study of Self-Tracking Data and Clinical Documentation in \nEndometriosis. Appl Clin Inform 11, 769–784 (2020). \n29. Ipek Ensari, Sharon Lipsky-Gorman, Emma N Horan, Suzanne Bakken, & Noémie Elhadad. Associations \nBetween Physical Exercise Patterns and Pain Symptoms in Individuals with Endometriosis: A Crosssectional \nmHealth-Based Investigation. BMJ Open 12, e059280 (2022). \n30. Ido Mick, Shay M. Freger, Jolanda van Keizerswaard, Mahsa Gholiof, & Mathew Leonardi. Comprehensive \nEndometriosis Care: A Modern Multimodal Approach For The Treatment of Pelvic Pain And Endometriosis. \nTher Adv Reprod Health 18, 1–23 (2024). \n31. Katherine Edgley, Philippa T. K. Saunders, Lucy H. R. Whitaker, Andrew W. Horne, & Athanasios Tsanas. \nInsights Into Endometriosis Symptom Trajectories and Assessment of Surgical Intervention Outcomes Using \nLongitudinal Actigraphy. npj Digital Medicine 8, 236 (2025). \n32. Adrienne Pichon, Iñigo Urteaga, Lena Mamykina, & Noémie Elhadad. Informing the Design of Individualized \nSelf-Management Regimens from the Human, Data, and Machine Learning Perspectives. ACM Transactions \non Computer-Human Interaction https://doi.org/https://dl.acm.org/doi/abs/10.1145/3717063 (2025) \ndoi:https://dl.acm.org/doi/abs/10.1145/3717063. \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n33. Pichon, A. et al. Divided We Stand: The Collaborative Work of Patients and Providers in an Enigmatic Chronic \nDisease. Proc ACM Hum Comput Interact 4, 261 (2021). \n34. Katherine Edgley, Andrew W. Horne, Philippa T.K. Saunders, & Athanasios Tsanas. Symptom Tracking in \nEndometriosis Using Digital Technologies: Knowns, Unknowns, and Future Prospects. Cell Reports Medicine \n4, 101192 (2023). \n35. McKillop, M., Mamykina, L. & Elhadad, N. Designing in the Dark: Eliciting Self-tracking Dimensions for \nUnderstanding Enigmatic Disease. in Proceedings of the 2018 CHI Conference on Human Factors in \nComputing Systems 1–15 (Association for Computing Machinery, New York, NY , USA, 2018). \ndoi:10.1145/3173574.3174139. \n36. Arnaud Fauconnier et al. Comparison of Patient- and Physician-Based Descriptions of Symptoms of \nEndometriosis: A Qualitative Study. Hum Reprod 28, 2686–2694 (2013). \n37. Solène Gouesbet et al. Patients’ Perspectives on How to Improve Endometriosis Care: A Large Qualitative \nStudy Within the ComPaRe-Endometriosis e-Cohort. Journal of Women’s Health 32, 463–470 (2023). \n38. Mathilde Bourdon et al. Investigating the Medical Journey of Endometriosis-Affected Women: Results From \nA Cross-Sectional Web-Based Survey (EndoVie) on 1,557 French Women. J Gynecol Obstet Hum Reprod 53, \n102708 (2024). \n39. Agneta Pettersson & Carina M. Bertero. How Women with Endometriosis Experience Health Care \nEncounters. Women’s Health Reports 529–542 (2020) doi:DOI: 10.1089/whr.2020.0099. \n40. Stella Bullo & Annalise Weckesser. Addressing Challenges in Endometriosis Pain Communication Between \nPatients and Doctors: The Role of Language. Front. Glob. Womens Health 2, (2021). \n41. Susanne Ilschner, Teresa Neeman, Melissa Parker, & Christine Phillips. Communicating Endometriosis Pain in \nFrance and Australia: An Interview Study. Front. Glob. Womens Health 3, (2022). \n42. Lisa Mikesell & Allyson C. Bontempo. Healthcare Providers’ Impact on the Care Experiences of Patients with \nEndometriosis: The Value of Trust. Health Communication 38, 1981–1993 (2023). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n43. Nastasja Robstad et al. Experiences of Pain Communication in Endometriosis: A Meta-Synthesis. Acta Obstet \nGynecol Scand 0, 1–16 (2024). \n44. Elisabeth Olliges et al. The Physical, Psychological, and Social Day-to-Day Experience of Women Living With \nEndometriosis Compared to Healthy Age-Matched Controls—A Mixed-Methods Study. Front. Glob. Womens \nHealth 2, (2021). \n45. Giulia Emily Cetera et al. “SO FAR AWAY” How Doctors Can Contribute to Making Endometriosis Hell on \nEarth. A Call for Humanistic Medicine and Empathetic Practice for Genuine Person-Centered Care. A \nNarrative Review. International Journal of Women’s Health 16, 273–287 (2024). \n46. Li, K. et al. Characterizing Physiological and Symptomatic Variation in Menstrual Cycles Using Self-Tracked \nMobile-Health Data. npj Digital Medicine 79, (2020). \n47. Urteaga, I., McKillop, M. & Elhadad, N. Learning Endometriosis Phenotypes from Patient-Generated Data. \nnpj Digital Medicine 3, 1–14 (2020). \n48. Karima Moumane & Ali Idri. Mobile Applications for Endometriosis Management Functionalities: Analysis \nand Potential. Scientific African 21, E01833 (2023). \n49. Camran Nezhat et al. Use of the Free Endometriosis Risk Advisor App as a Non-Invasive Screening Test for \nEndometriosis in Patients with Chronic Pelvic Pain and/or Unexplained Infertility. J. Clin. Med 12, 5234 \n(2023). \n50. Esther van Barneveld et al. Real-time Symptom Assessment in Patients With Endometriosis: Psychometric \nEvaluation of an Electronic Patient-Reported Outcome Measure, Based on the Experience Sampling Method. \n7, e29480 (2023). \n51. Marco Richard Zugaj, Ariane Germeyer, Karina Kranz, Andrea Züger, & Jens Keßler. Experiences of Patients \nWith Endometriosis With A Digital Health Application: A Qualitative Analysis. Archives of Gynecology and \nObstetrics 310, 2253–2263 (2024). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n52. Aidan P Wickham et al. Exploring Self-Reported Symptoms for Developing and Evaluating Digital Symptom \nCheckers for Polycystic Ovarian Syndrome, Endometriosis, and Uterine Fibroids: Exploratory Survey Study. \nJMIR 8, e65469 (2024). \n53. Abhishek Pratap et al. Indicators of Retention in Remote Digital Health Studies: A Cross-Study Evaluation of \n100,000 Participants. npj Digit. Med 3, 21–30 (2020). \n54. Elliot G. Brown, Louise Wood, & Sue Wood. The Medical Dictionary for Regulatory Activities (MedDRA). Drug \nSafety 20, 109–117 (1999). \n55. MedDRA® the Medical Dictionary for Regulatory Activities terminology is the international medical \nterminology developed under the auspices of the International Council for Harmonisation of Technical \nRequirements for Pharmaceuticals for Human Use (ICH), https://www.meddra.org/faq. \n56. J. Lin. Divergence Measures Based On The Shannon Entropy. IEEE Transactions on Information Theory 37, \n145–151 (2006). \n57. M. Aziz, M. A. Beaton, M. A. Aziz, J. Opoku-Anane, & N. Elhadad. Endometriosis and AutoEmmunity: A \nLargescale Case-Control Study of endometriosis and 10 Distinct Autoimmune Diseases. npj Women’s Health \n3, 36–42 (2025). \n58. A. Fauconnier & C. Chapron. Endometriosis and Pelvic Pain: Epidemiological Evidence of the Relationship \nand Implications. 11, 595–606 (2005). \n59. Karen Ballard, Hazel Lane, Gernot Hudelist, Saikat Banerjee, & Jeremy Wright. Can Specific Pain Symptoms \nHelp in the Diagnosis of Endometriosis? A Cohort Study of Women with Chronic Pelvic Pain. Fertility and \nSterility 94, 20–27 (2010). \n60. Hedyeh Riazi et al. Clinical Diagnosis of Pelvic Endometriosis: A Scoping Review. BMC Women’s Health 15, \n(2015). \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n61. Nikolaos V. Apostolopoulos, Krystallenia I. Alexandraki, Anwen Gorry, & Adeyemi Coker. Association Between \nChronic Pelvic Pain Symptoms and the Presence of Endometriosis. Arch Gynecol Obstet 293, 439–445 \n(2016). \n62. Ahmed M. Soliman, Karin S. Coyne, Erica Zaiser, Jane Castelli-Haley, & Mahesh J. Fuldeore. The Burden of \nEndometriosis Symptoms on Healthrelated Quality of Life in Women in the United States: A Cross-Sectional \nStudy. Journal of Psychosomatic Obstetrics & Gynecology 38, 238–248 (2017). \n63. McKillop, M., Mamykina, L. & Elhadad, N. Designing in the Dark: Eliciting Self-Tracking Dimensions for \nUnderstanding Enigmatic Disease. CHI 565, 1–15 (2018). \n64. https://endoisrael.org/clinics/. \n65. Georgina Jones, Georgina Jones, & Georgina Jones. Evaluating The Responsiveness Of The Endometriosis \nHealth Profile Questionnaire: The EHP-30. Quality of Life Research 13, 705–713 (2004). \n66. K.E. Hansen et al. Health-Related Quality Of Life In Women With Endometriosis: Psychometric Validation Of \nThe Endometriosis Health Profile 30 Questionnaire Using Confirmatory Factor Analysis. Human \nReproduction Open 2022, 1–11 (2022). \n67. D. M. Endres & J. E. Schindelin. A New Metric For Probability Distributions. IEEE Transactions on Information \nTheory 49, 1858–1860 (2006). \n68. Yoav Benjamini & Yosef Hochberg. Controlling the False Discovery Rate: A Practical and Powerful Approach \nto Multiple Testing. Journal of the Royal Statistical Society. Series B (Methodological) 57, 289–300 (1995). \n69. Ary L. Goldberger et al. PhysioBank, PhysioToolkit, and PhysioNet: Components of A New Research Resource \nFor Complex Physiologic Signals. Circulation 101, e215–e220 (2000). \n70. Zelovich, T et al. KI EndoLIST: Endometriosis Longitudinal Individualized Symptoms Tracking Dataset. \nhttps://doi.org/https://doi.org/10.13026/k99q-fm63 (2026). \n \n \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n  \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nTable 1: Study cohort characteristics \nN* 34 Fertility problem†   \nAge* [years]  Yes 2 (6%) \nWhen joining the study 33 [30.25, 36.75] No 6 (18%) \nAt first period 12 [11.12, 13.75] Unknown 26 (76%) \nAt first endometriosis symptom 14 [12.75, 17] Adenomyosis†  \nAt endometriosis diagnosis 29 [24, 31] Yes 17 (50%) \n  No 13 (38%) \nBMI* [kg/m2] 21.89 [20.36, 23.41] Don't know 4 (12%) \nFamily history†  Diagnostic surgery†  \nMom 5 (15%) No 28 (82%) \nSister 2 (6%) Yes 6 (18%) \nAunt 5 (15%) One surgery 5 (15%) \nCousin 2 (6%) >1 surgery 1 (3%) \nGrandmother 2 (6%) Menstrual regularity†  \nGrandma's cousin 1 (3%) Regular 31 (91%) \nUnknown 22 (65%) Irregular 3 (9%) \n* Median [IQR]; † Number of participants (% of the study cohort). BMI: Body mass index. \n  \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \nTable 2:  Study Characteristics. \nTracked symptoms^,*    Tracked cycles    \nNumber of app symptoms  24.5 [19.25, 36.5] Number of cycles*  5 [3, 7]  \nNumber of LLT entities  28 [20.2, 40.8] Less than 3†  6 (18%)  \nNumber of PT entities  25.5 [20, 36.8] Three or more†    28 (82%)  \nNumber of HLT entities  27.5 [21.2, 38.8] Cycle length*  28.5 [27, 31.1]  \nNumber of HLGT entities  19 [16, 25]   \nNumber of SOC entities  10 [9, 12] Study incompletion reason†    \nTracked days*   Pregnancy  4 (12%)  \nTotal days in the study  134 [93.5, 218.8] Personal (layoffs, divorce)  2 (6%)  \nEngagement [%]#  95.3 [76.8, 97.4] War-related reasons  1 (3%)  \n^ Symptoms mapped to MedDRA (excluding General emotional condition, General physical condition, \nOvulation test values, Bleeding, medications and well -being related entities); # Percent of days with \nreports for at least half of the symptoms; * Median [IQR]; † Number of participants (% of the study \ncohort).  \nTable 3: Frequencies of SOCs derived from symptoms tracked by participants.  \nSOC #Participants (%) #LLT Top LLTs* (#Participants†) \nGastrointestinal disorders 34 (100%) 29 \n• Pelvic pain female (27) \n• Diarrhea (25) \n• Constipation (24) \n• Defecation urge painful (24) \n• Dyschezia (24) \nMusculoskeletal and connective \ntissue disorders 34 (100%) 28 \n• Low back pain (28) \n• Defecation urge painful (24) \n• Scapula pain (15) \nReproductive system and breast \ndisorders 34 (100%) 18 \n• Endometriosis related pain (27) \n• Pelvic pain female (27) \n• Breast tenderness (22) \n• Painful periods (22) \nGeneral disorders and \nadministration site conditions 34 (100%) 14 \n• Fatigue (27) \n• Increased appetite (21) \n• Decreased appetite (19) \nNervous system disorders 33 (97%) 26 \n• Headache (18) \n• Brain fog (15) \n• Dizziness (15) \nRenal and urinary disorders 29 (85%) 8 \n• Pelvic pain female (27) \n• Incomplete urination (13) \n• Urination frequency of (11) \nVascular disorders 28 (82%) 9 \n• Dizziness (15) \n• Migraine (11) \n• Hot flushes (9) \nMetabolism and nutrition disorders 26 (76%) 4 \n• Increased appetite (21) \n• Decreased appetite (19) \n• Food craving (15) \nPsychiatric disorders 22 (65%) 10 \n• Anger (15) \n• Brain fog (15) \n• Anxiety (8) \n\nARTICLE IN PRESS\nARTICLE IN PRESS\n \n \n \n• Depression (8) \nRespiratory, thoracic and \nmediastinal disorders 20 (59%) 12 \n• Smell alteration (7) \n• Chest pain (6) \n• Rib pain (4) \n• Shortness of breath (4) \nCardiac disorders 18 (53%) 6 \n• Dizziness (15) \n• Chest pain (6) \n• Shortness of breath (4) \nSkin and subcutaneous tissue \ndisorders 15 (44%) 11 \n• acne (11) \n• Touch sensitivity increased (4) \n• Hair Loss (2) \n• Numbness in hands, forearms, elbows (2) \n• Vaginal itching (2) \nImmune system disorders 9 (26%) 6 \n• Multiple allergies (5) \n• Nasal allergy (3) \n• Asthma (2) \nInfections and infestations 8 (24%) 5 \n• Sinusitis (3) \n• Eye infection (2) \n• Gum infection (2) \nEye disorders 6 (18%) 7 \n• Blurred vision (2) \n• Eye infection (2) \n• Eye allergy (1) \n• Eye pain (1) \n• Sensation of pressure in eye (1) \n• Sjogren's (1) \n• Visual disturbances (1) \nEar and labyrinth disorders 4 (12%) 2 • Sound sensitivity increased (3) \n• Ear pain (2) \nEndocrine disorders 1 (3%) 1 • Small fiber neuropathy (1) \n \n*In cases where multiple LLTs share the same frequency as the third most common term, all tied LLTs \nare included; †number of participants who tracked each LLT.","source_license":"public-domain-us","license_restricted":false}