Applying Machine Learning to Understanding Engagement Patterns
The transformative power of the current health crisis has been expansive and affected all fields and industries. One of the major shifts the situation has brought with it is the quickened pace of adoption of technology and virtual solutions, opening society up to optimized productivity and processes in the new digital economy. These virtual solutions are normally designed with people in mind with the purpose to enhance what we do while remaining person-centric. Keeping the individual at the center of technology allows its application in fields normally reserved for exclusively in-person interactions, like psychology and mental health therapies.
These challenging times have brought with them mental health strains most have not been exposed to before. A quickly emerging remote solution for mental health is internet-based cognitive behavioral therapy (iCBT). Cognitive behavior therapy (CBT) is a psycho-social intervention with the purpose of improving mental health. Given the challenge of social distancing during COVID-19, internet-based solutions for CBT is a safe option for all involved parties.
There has been extensive study on patient engagement and digital solutions. The research concludes that there are measurable and reliable positive outcomes drawn from iCBT. For example, one such study concludes that: “Cognitive behavioral therapy delivered over the Internet leads to immediate and sustained reduction in depressive symptoms; thus, it may be a good treatment modality for individuals unable or unwilling to access traditional face-to-face therapy” (Efficacy of cognitive behavioral therapy, n.d.).
Applying Machine Learning to Understanding Engagement Patterns
At Mozzaz we are always enthusiastic about incorporating the latest data science and machine learning (ML) applications into our existing solutions. This is why we were particularly interested with a recent study by Chien I, Enrique A, Palacios J, et al., titled ‘A Machine Learning Approach to Understanding Patterns of Engagement With Internet-Delivered Mental Health Interventions’. The study applied ML models to identify patterns of patient behavior that could be used to tailor different types of digital engagement strategies.
The study was able to identify five classes of users that was based on their interaction patterns with the iCBT programs. These classes showed that clinical outcomes are not only linked to the time spent on the program or activity, but could be affected by other factors, such as: “Class 4 engaged more in goal-based activities and mood tracking and accessed many core modules, whereas class 5 participants were less likely to access core modules, but used relaxation and mindfulness tools.” (Chien I, Enrique A, Palacios J, et al., 2020). Both Class 4 & Class 5 respectively saw a 58.8% and 66.9% reliable improvement in mental health (Chien I, Enrique A, Palacios J, et al., 2020). With more available data ML analysis can further identify and categorize more use case patterns in order to individually customize internet-based therapeutic programs.
Furthermore, the reliable improvements for patients across the different classes followed a positive trend “from class 1 (39.5%) to class 5 (66.9%)” (Chien I, Enrique A, Palacios J, et al., 2020). This trend suggests that if more classes can be identified in order to individualize the iCBT program or activity, there would be more room for even stronger reliable improvements in mental health for the patient.
As briefly mentioned, the study found that positive outcomes of improved mental health were not correlated with the amount of use time or frequency of use of an iCBT application, but rather depended on the type of interaction patterns with the digital therapy in question. Leveraging this knowledge and focusing on the identification of use patterns changes the fundamental approach to iCBT from one-size-fits all to the need for focus on customizable iCBT programs in real-time.
References
Chien I, Enrique A, Palacios J, et al. A Machine Learning Approach to Understanding Patterns of Engagement With Internet-Delivered Mental Health Interventions. JAMA Netw Open. 2020;3(7):e2010791. doi:10.1001/jamanetworkopen.2020.10791
Efficacy of cognitive behavioral therapy delivered over the Internet for depressive symptoms: A systematic review and meta-analysis, pubmed.ncbi.nlm.nih.gov/28696153/
Internet-based vs. face-to-face cognitive behavior therapy for psychiatric and somatic disorders: an updated systematic review and meta-analysis www.tandfonline.com/doi/full/10.1080/16506073.2017.1401115
Psycho-social intervention with the purpose of improving mental health, wikipedia.org/wiki/Cognitive_behavioral_therapy
Sztein DM, Koransky CE, Fegan L, Himelhoch S. Efficacy of cognitive behavioural therapy delivered over the Internet for depressive symptoms: A systematic review and meta-analysis. J Telemed Telecare. 2018;24(8):527-539. doi:10.1177/1357633X17717402
