Can AI help us to understand Endometriosis?

In this article, I would like to discuss the role of period tracking apps to diagnose and understand endometriosis. I argue that the path to diagnose endometriosis depends on many different visual or even “material” evidences. I will reflect on my research about gynecologists’ understanding of endometriosis to discuss how data conducted by period tracking apps could support medical diagnosing practices.

Endometriosis is an estrogen-dependent gynecological condition that can occur in people who have a uterus and are in their reproductive life span (van der Zanden et al. 2018). A characteristic of the condition is the existence of inflammatory endometrial-like tissues which grow outside the uterus in different intensities (Dunselman et al. 2014, p. 401). It takes on average 6.7 years to diagnose the chronic disease (van der Zanden et al. 2018). Often, endometriosis is misdiagnosed or normalized during medical encounters despite the intensity of symptoms in some cases (Grundström et al. 2016). Its diagnostic delay has many reasons, one of them might be the enigmatic character of endometriosis itself. Enigmatic diseases challenge medical diagnosing practices due to their high level of individuality. Since endometriosis is a cycle depending on chronic disease, all its symptoms may occur in accordance with the cycle. Thus, it is very challenging for gynecologists to draw the line between dysmenorrhea and mild endometriosis. Since endometriosis’ intensity does not match the biological symptoms it causes, diagnosing the disease is even more difficult (Soliman et al. 2017). Some patients would not feel anything and have deep infiltrative endometriosis, others would suffer from tremendous pain and pathologists can only find tiny endometrial tissues growing outside of the uterus. There is an endometriosis questionnaire to identify mild or severe endometriosis, the WERF EPHect (Vitonis et al., 2014 in Urteaga et al. 2020, p. 3). However, my research shows that it is extremely difficult to identify endometriosis in contrast to symptoms of the cycle despite categorizations and questionnaires. Pain, bleeding and the cycle itself are highly individual phenomena which are difficult to standardize. Doctors rely on patients’ reports to diagnose it. Since the disease is so diverse, patients cannot be sure if they really suffer from the disease or not.

In general, gynaecologists indicate an endometriosis diagnosis if they see many of the typical symptoms and if they can visually verify them by doing ultrasound or a laparoscopy. Ultrasound cannot deliver 100% certainty about endometriosis’ existence, a laparoscopy is still considered as the gold standard of diagnosing practice (DGGG et al. 2020; Taylor et al. 2018). A suspected diagnosis can be made if doctors think the patient suffers from endometriosis without having it verified by a pathological finding. The latter case is not recorded as official endometriosis diagnosis in the insurance records. The number of endometriosis’s diagnosis can therefore be expected to be higher than it really is. Very severe symptoms lead to a diagnosis or suspected diagnosis of endometriosis more easily. I argue that solid and verified numbers about endometriosis cases can increase the willingness to invest in research about endometriosis as well as gynecologists’ general awareness about the disease.

What happens when gynaecologists are confronted with patients suffering from symptoms indicating endometriosis? Doctors aim at enhancing the overall well-being of patients. This means they likely prescribe medication like the birth-control pill to stop the cycle and thus ease the symptoms. In other words: doctors apply technologies to stop the cycle — which helps patients a lot — however, as soon as the cycle stops, typical signs of endometriosis also disappear in the majority of cases. If there was not a clear diagnosis or at least a clear suspicion for endometriosis, the disease is likely to be forgotten and will not be uncovered until symptoms become severe in the future. This will often happen when patients stop taking the pill.

Not only patients benefit from the pill, also doctor’s offices benefit from its prescription. On average, doctors have 8 minutes to treat patients. If they need more time, they can treat less patients ion a day, thus they earn less money. This brings gynaecologists into a financial dilemma. It takes time to diagnose enigmatic diseases — especially endometriosis — but doctors do not have this time. However, the more prescriptions they sign the more money they will get. The result is that a German patient who asks for a new anti-baby pill prescription every three month is extremely valuable for doctors’ offices to meet their monthly costs. This system was not made by German gynaecologists, but by the German insurance system. In other words, there are two reasons why patients who suffer from symptoms indicating endometriosis will likely receive the birth-control pill: first, the pill helps to ease the symptoms, second, doctors have a financial benefit from it which they also need to pay for their doctor’s office. Thus, this approach helps patients and doctors but it also subverts diagnosing the disease, because typical symptoms are stopped.

Different studies discuss how gender misconceptions about diseases like endometriosis make it even more challenging for patients to state their complaints if they are not related to infertility problems or if patients are unmarried (Young et al. 2019, p. 7; Whelan 2007, p. 958; Seear 2014). However, this is not the rule of thumb and many gynaecologists appear to be extremely dedicated and empathic about the suffering of patients. Still, the suffering of chronic diseases is shaped by culture, therein forming control, stigma and autonomy (Charmaz 2010). The (invisible) suffering caused by chronic illnesses and diseases, like endometriosis, thus becomes a matter of social discreditation and medical uncertainty (ibid.; Whelan 2007). Furthermore, menstruation was a taboo topic until recently in societies of the global north and is in some cultures still today (Culley et al. 2011).

Recently, numerous period tracking apps have entered the market. The apps offer users to track many different symptoms of the period like pain, cramps, tenderness of breasts, skin appearance and also lifestyle aspects like sleep quality, sexual activity etc. The apps gather data from self-tracking individuals that are in turn rewarded with knowledge about their individual cycle (Kressbach 2021, p.2). App developers promise to educate users about their menstrual health or support in contraception (ibid.). Some period tracking apps have cooperation with academic institutes to conduct women’s health centric medical research (see Wheeler et al. 2015).

Period tracking apps are self-tracking health apps that encourage users to track and engage more intensely with their own bodily functions. Using big data as an epistemological authority, these apps claim to provide scientific evidence about individual health conditions (Kressbach 2021, p. 16). A predictive algorithm estimates cycle length, up-coming periods, the fertility window and mood changes like PMS (Wheeler et al. 2015).

There are also digital health apps particularly designed for endometriosis that focus on specific symptoms characterizing the disease (see “Phendo” or “Bayer Endo Diary”). Research about self-tracking applications for chronic diseases like endometriosis suggests that it can inform evidence-based care and support finding evidence for poorly understood chronic diseases (Ensari et al. 2020, p. 780). The proper documentation of symptoms related to e. g. sleep quality can illuminate individual symptomatic pictures and thus enhance deeper understanding about the still enigmatic characteristics of endometriosis (ibid.). Despite perceived burdens to track bodily patterns on a daily basis, allocated data can complement clinical documentation and emphasize the benefit of patients’ engagement in treating and researching endometriosis (Ensari et al. 2020, p. 781). Furthermore, health tracking apps designed for users with chronic diseases are perceived as empowering to manage symptoms independently ( Birkhoff & Smeltzer 2017, p.4).

Does digital, quantified information about patients’ suffering reduce uncertainty about patients’ suffering from mild and severe symptoms? When reflecting on my research about German gynaecologists dealing with endometriosis, I believe that doctors seem to need some kind of “material” evidence about patients’ suffering if symptoms are rather mild or if a pathological verification (endometrial tissues) cannot be seen. A visual verification of symptoms even helps gynaecologists cope with medical uncertainty. So, I argue that data about frequency and location of pain, intensity, amount of bleeding etc. could support diagnosing practices as well. Big data analytics is suspected to be a new epistemology for understanding uncertain worldly patterns (Pink et. al. 2016). Likewise, Big data’s epistemological promise to assist in medical decision making is already implicitly mentioned by period tracking apps. Indeed, my research has shown that gynaecologists seem to need some kind of “material” verification to settle their suspicion. Even though this “material” verification stands in sharp contrast to the numerous severe and highly individual symptoms patients suffer from. Since diseases like endometriosis are highly individual, I believe that it can be extremely helpful to see the data records of a single patient to raise the awareness about a potential endometriosis disease and to finalize a suspicion.

Previously research shows that using an endometriosis period tracking app can help defining endometriosis digital phenotypes between mild endometriosis and severe endometriosis (see Urteaga et al. 2020). Maybe this could also allow doctors to draw more confidently the line between mild endometriosis and dysmenorrhea? Phenotyping digitally as well as biologically endometriosis remains as an open research question (ibid.) Apps seem to have a real chance to increase the knowledge about endometriosis.

However, knowledge production is embedded in socio-cultural and socio-technical contingencies from the past to the present. Menstrual irregularities exist as long as individuals menstruate and also records of diseases like endometriosis reach far back into the past (Nezhat et. al. 2012). Big data offers a new way to see worldly things differently, but this difference also depends on how people perceive their worlds (Pink et al. 2016). Endometriosis has become a political issue since it is a chronic disease that relates to all aspects in human life. This anchors the disease in the discourse of gender discrimination. Big Data will deliver “new knowledge” but can it change how individuals interpret this knowledge? If individuals do not see “Small Data”, how can big data tell a different story (Pink et al. 2016, p. 2)? Research should investigate how doctors would relate big health data to individual cases and patterns of chronic diseases.

Socio-technical assemblages will also influence how knowledge about endometriosis is translated into digital health data. Often doctors recommend taking the birth-control pill or other hormonal therapy to ease the severe symptoms, before an official endometriosis diagnosis was made. So, if patients are on the pill, to what extent would their data help to define an endometriosis phenotype or to become suspicious about endometriosis? Furthermore, to what extent can the data support medical diagnosing practices and decision making? If patients take the birth-control pill and track their individual symptoms, aren’t there two competing technologies at work that subvert each other’s benefits? What impact would this have on the predictive algorithm in the context of a potential endometriosis? Users can inform the app about their contraception methods, however, I argue that reactions by the pill as well as by endometriosis are highly individual. Therefore, it would be very interesting to study how predictive algorithms make sense out of these individual data packages.

Nevertheless, digital technologies offer patients to engage and cope with their symptoms. And it seems likely that this data can also enhance medical practices concerning endometriosis. More data is definitely needed to fully understand the disease and also its digital digestion.

References

“Bayer Endo Diary.” 2021. https://www.viziofly.com/portfolio/bayer-endo-diary/.

Birkhoff, Susan D., and Suzanne C. Smeltzer. 2017. “Perceptions of Smartphone User-Centered Mobile Health Tracking Apps Across Various Chronic Illness Populations: An Integrative Review.” Journal of Nursing Scholarship 49 (4): 371–78. https://doi.org/10.1111/jnu.12298.

Charmaz, K. (2010). Chronic Illness. In New Blackwell Companion to Medical Sociology (pp. 312–333). London: Blackwell Publishing Ltd. https://doi.org/10.1017/CBO9781107415324.004

Culley, Lorraine, Irena Papadopoulos, Elaine Denny, and Patricia Apenteng. 2011. “From Womanhood to Endometriosis: Findings from Focus Groups with Women from Different Ethnic Groups.” Diversity & Equality in Health and Care 8 (3). https://diversityhealthcare.imedpub.com/abstract/from-womanhood-to-endometriosis-findings-from-focus-groups-with-women-from-different-ethnic-groups-1879.html.

Deutsche Gesellschaft für Gynäkologie (DGGG), Österreichische Gesellschaft für Gynäkologie und Geburtshilfe (OEGGG), and Schweizerische Gesellschaft für Gynäkologie und Geburtshilfe (SGGG). 2020. “Leitlinienprogramm. Diagnostik Und Therapie Der Endometriose.,” 163. https://www.awmf.org/uploads/tx_szleitlinien/015- 045l_S2k_Diagnostik_Therapie_Endometriose_2020–09.pdf.

Dunselman, G.A.J., N. Vermeulen, C. Becker, C. Calhaz-Jorge, T. D’Hooghe, B. De Bie, O. Heikinheimo, et al. 2014. “ESHRE Guideline: Management of Women with Endometriosis †.” Human Reproduction 29 (3): 400–412. https://doi.org/10.1093/humrep/det457.

Ensari, Ipek, Adrienne Pichon, Sharon Lipsky-Gorman, Suzanne Bakken, and Noémie Elhadad. 2020. “Augmenting the Clinical Data Sources for Enigmatic Diseases: A Cross-Sectional Study of Self-Tracking Data and Clinical Documentation in Endometriosis.” Applied Clinical Informatics 11 (05): 769–84. https://doi.org/10.1055/s-0040-1718755.

Grundström, Hanna, Preben Kjølhede, Carina Berterö, and Siw Alehagen. 2016. “‘A Challenge’ — Healthcare Professionals’ Experiences When Meeting Women with Symptoms That Might Indicate Endometriosis.” Sexual & Reproductive Healthcare: Official Journal of the Swedish Association of Midwives 7 (March): 65–69. https://doi.org/10.1016/j.srhc.2015.11.003.

Kressbach, Mikki. 2021. “Period Hacks: Menstruating in the Big Data Paradigm.” Television & New Media 22 (3): 241–61. https://doi.org/10.1177/1527476419886389.

Nezhat, Camran, Farr Nezhat, and Ceana Nezhat. 2012. “Endometriosis: Ancient Disease, Ancient Treatments.” Fertility and Sterility 98 (6): S1–62. https://doi.org/10.1016/j.fertnstert.2012.08.001.

“Phendo.” 2016. http://citizenendo.org/phendo/.

Pink, Sarah, Deborah Lupton, Martin Berg, Paul Dourish, Adrian Dyer, Vaike Fors, Edgar Cruz, et al. 2016. DATA ETHNOGRAPHIES (1): Personal Data in an Uncertain World. https://doi.org/10.13140/RG.2.2.27713.76643.

Seear, Kate. 2014. The Makings of a Modern Epidemic: Endometriosis, Gender and Politics. Ashgate Publishing Limited. https://research.monash.edu/en/publications/the-makings-of-a-modern-epidemic-endometriosis-gender-and-politic.

Soliman, Ahmed M., Karin S. Coyne, Katharine S. Gries, Jane Castelli-Haley, Michael C. Snabes, and Eric S. Surrey. 2017. “The Effect of Endometriosis Symptoms on Absenteeism and Presenteeism in the Workplace and at Home.” Journal of Managed Care & Specialty Pharmacy 23 (7): 745–54. https://doi.org/10.18553/jmcp.2017.23.7.745.

Taylor, Hugh S., G. David Adamson, Michael P. Diamond, Steven R. Goldstein, Andrew W. Horne, Stacey A. Missmer, Michael C. Snabes, Eric Surrey, and Robert N. Taylor. 2018. “An Evidence-Based Approach to Assessing Surgical versus Clinical Diagnosis of Symptomatic Endometriosis.” International Journal of Gynecology & Obstetrics 142 (2): 131–42. https://doi.org/10.1002/ijgo.12521.

Urteaga, Iñigo, Mollie McKillop, and Noémie Elhadad. 2020. “Learning Endometriosis Phenotypes from Patient-Generated Data.” Npj Digital Medicine 3 (1): 88. https://doi.org/10.1038/s41746-020-0292-9.

van der Zanden, M., M.W.J. Arens, D.D.M. Braat, W.L.M. Nelen, and A.W. Nap. 2018. “Gynaecologists’ View on Diagnostic Delay and Care Performance in Endometriosis in the Netherlands.” Reproductive BioMedicine Online 37 (6): 761–68. https://doi.org/10.1016/j.rbmo.2018.09.006.

Vitonis, Allison F., Katy Vincent, Nilufer Rahmioglu, Amelie Fassbender, Germaine M. Buck Louis, Lone Hummelshoj, Linda C. Giudice, et al. 2014. “World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project: II. Clinical and Covariate Phenotype Data Collection in Endometriosis Research.” Fertility and Sterility 102 (5): 1223–32. https://doi.org/10.1016/j.fertnstert.2014.07.1244.

Wheeler, Marija Vlajić, Vedrana Högqvist Tabor, Kayleigh Teel, and Mike LaVigne. 2015. “The Science of Your Cycle: Evidence-Based App Design.” December 10, 2015. https://helloclue.com/articles/about-clue/science-your-cycle-evidence-based-app-design.

Whelan, Emma. 2007. “‘No One Agrees except for Those of Us Who Have It’: Endometriosis Patients as an Epistemological Community.” Sociology of Health & Illness 29 (7): 957–82. https://doi.org/10.1111/j.1467-9566.2007.01024.x.

Young, Kate, Jane Fisher, and Maggie Kirkman. 2019. “‘Do Mad People Get Endo or Does Endo Make You Mad?’: Clinicians’ Discursive Constructions of Medicine and Women with Endometriosis.” Feminism & Psychology 29 (3): 337–56. https://doi.org/10.1177/0959353518815704.