Increasing the value of digital phenotyping through reducing missingness: a retrospective review and analysis of prior studies

Abstract

BackgroundDigital phenotyping methods present a scalable tool to realise the potential of personalised medicine. But underlying this potential is the need for digital phenotyping data to represent accurate and precise health measurements.ObjectiveTo assess the impact of population, clinical, research and technological factors on the digital phenotyping data quality as measured by rates of missing digital phenotyping data.MethodsThis study analyses retrospective cohorts of mindLAMP smartphone application digital phenotyping studies run at Beth Israel Deaconess Medical Center between May 2019 and March 2022 involving 1178 participants (studies of college students, people with schizophrenia and people with depression/anxiety). With this large combined data set, we report on the impact of sampling frequency, active engagement with the application, phone type (Android vs Apple), gender and study protocol features on missingness/data quality.FindingsMissingness from sensors in digital phenotyping is related to active user engagement with the application. After 3 days of no engagement, there was a 19% decrease in average data coverage for both Global Positioning System and accelerometer. Data sets with high degrees of missingness can generate incorrect behavioural features that may lead to faulty clinical interpretations.ConclusionsDigital phenotyping data quality requires ongoing technical and protocol efforts to minimise missingness. Adding run-in periods, education with hands-on support and tools to easily monitor data coverage are all productive strategies studies can use today.Clinical implicationsWhile it is feasible to capture digital phenotyping data from diverse populations, clinicians should consider the degree of missingness in the data before using them for clinical decision-making.

Digital Tools to Facilitate the Detection and Treatment of Bipolar Disorder: Key Developments and Future Directions

“…User comfort in sharing data differs depending on the data type (eg, users are more comfortable sharing health data than personal data such as location, communication logs, and social activity) and the recipient (eg, users have greater comfort sharing data directly with clinicians than having this entered into their electronic health record), and this may impact willingness to use digital phenotyping platforms [ 21 ]. As user engagement is essential for the success of any digital phenotyping tools [ 22 ], it is necessary to account for discrepancies in access, equity, and distribution of resources. Finally, more in-depth longitudinal studies are required to ascertain the relationship between biomarkers and long-term outcomes of health and well-being.…”

Digital Mental Health for Schizophrenia and Other Severe Mental Illnesses: An International Consensus on Current Challenges and Potential Solutions

“…4. Research should place an equal emphasis on social and population factors [135] as well as biological and social factors in the etiology and maintenance of symptoms and risk in SMI. 5. Successful implementation and application of digital mental health in real-world clinical settings will require new and evolving collaborations between academics, clinicians, people with lived experience of SMI, industrial designers, software developers, and regulatory specialists.…”

“…The pandemic showed that a shift to digital platforms to deliver synchronous mental health services could be rapidly implemented [ 2 ] and can be an acceptable format for many clinicians, patients, and carers [ 3 , 4 ]. However, asynchronous digital approaches (eg, measuring symptoms using digital phenotyping or ecological momentary assessment or providing partially automated therapies using digital platforms) show even more potential to increase capacity and outcomes [ 5 ]. These innovations allow patients to undertake a variety of clinically relevant tasks (such as self-monitoring or therapy tasks) outside the in-person clinical encounter.…”

A Template and Tutorial for Preregistering Studies Using Passive Smartphone Measures

“…This determination can be made through theoretical reasoning or by conducting a pilot study (Velozo et al, 2022). A widely used guideline suggests that the sampling frequency should be at least twice the frequency of the smallest expected fluctuation in the variable of interest (Bogdan, 2009), especially when considering that sampling frequency may be less than anticipated due to pushback from the underlying operating system (Currey & Torous, 2023;Niemeijer et al, 2023). Nevertheless, when deciding on a sampling frequency, the researcher should consider practical constraints, such as battery drain, which could reduce participant retention.…”