Machine Learning for Predicting Adherence to Internet-Delivered Psychotherapy for Symptoms of Depression and Anxiety after Myocardial Infarction: Insights from the U-CARE Heart Trial (Preprint)

Abstract

BACKGROUND BACKGROUND: Low adherence to recommended treatments is a multifactorial problem in rehabilitation for patients with myocardial infarction (MI). In a nationwide trial of internet-delivered cognitive behavior therapy (iCBT) for the high-risk subgroup of patients with MI also reporting symptoms of anxiety, depression, or both (MI-ANXDEP), adherence was low. Since low adherence to psychotherapy leads to a waste of therapeutic resources and risky treatment abortion in MI-ANXDEP patients, identifying early predictors for adherence is potentially valuable for effective targeted care. OBJECTIVE Applied predictive modelling with supervised machine learning to investigate both established and novel predictors for iCBT adherence in MI-ANXDEP patients. METHODS Data were from 90 MI-ANXDEP patients recruited from 25 hospitals in Sweden and randomized to treatment in the iCBT trial U-CARE Heart. Time-point of prediction was at completion of the first homework assignment. Adherence was defined as having completed at least the first two homework assignments within the 14-week treatment period. A supervised machine learning procedure was applied to identify the most potent predictors for adherence available at the first treatment session from a range of demographic, clinical, psychometric, and linguistic predictors. The internal binary classifier was a random forest model within a 3x10-fold cross-validated recursive feature elimination (RFE) resampling, which selected the final predictor subset which best differentiated adherers versus non-adherers. RESULTS RESULTS: Patient mean age was 58.4 (9.4) years, 62% (56/90) were men, and 48% (43/90) were adherent. Out of the 34 potential predictors for adherence, RFE selected an optimal subset of 56% (19/34) (Accuracy 0.64, 95% CI 0.61-0.68, P < 0.01). The strongest predictors for adherence were in order of importance (1) self-assessed cardiac-related fear, (2) sex, and (3) the number of words the patient used to answer the first homework assignment. CONCLUSIONS CONCLUSIONS: Adherence to iCBT for MI-ANXDEP patients was best predicted by cardiac-related fear and sex, consistent with previous research, but also by novel linguistic predictors from written patient behavior which conceivably indicate verbal ability or therapeutic alliance. Future research should investigate potential causal mechanisms, seek to determine what underlying constructs the linguistic predictors tap into, and whether these findings replicate for other interventions, outside of Sweden, in larger samples, and for patients with other conditions whom are offered iCBT. CLINICALTRIAL TRIAL REGISTRATION: ClinicalTrials.gov NCT01504191; https://clinicaltrials.gov/ct2/show/NCT01504191 (Archived at Webcite at http://www.webcitation.org/6xWWSEQ22)

Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review

“…The identified studies (n=29) were mostly conducted in the USA (n=10), Sweden (n=6), and the UK (n=6). The studies included a diverse sample of adults (n=22), college students and young adults (n=4), anonymous users (n=2), and adolescents (n=1) ( 28 – 33 , 41 , 73 – 94 ). Eight studies reported on the ethnicity of the participants ( 30 , 32 , 41 , 74 , 76 , 77 , 80 , 83 ), and in most of the studies between 31% and 91% of the participants identified as White.…”

“…About half of the identified studies (41.4%) focused on the implementation and analysis of a variety of AI prediction algorithms for different outcomes in online mental healthcare, including treatment responses and symptom remission, treatment dropouts, and treatment engagement and adherence ( 28 – 32 , 87 – 89 , 91 – 94 ). The identified studies applied ML algorithms such as: DL, and ANN algorithms to process large amounts of treatment data to identify helpful prediction factors.…”

“…The specific AI data analysis approaches used by the relevant studies and an assessment of study elements following research guidelines and standards for ML studies (i.e., adequacy of the data for the intended outcomes, model training and fine-tuning, features analyzed, validation, interpretability, and code and data availability) are presented in Table 1 . All ML studies [n=15 ( 28 – 33 , 85 – 89 , 91 – 94 )] presented a comprehensive description of their data, justification for the implementation of ML in their analysis, and a description of the features analyzed. 80% of studies reported the data or approach used to train their ML algorithms ( 28 – 30 , 32 , 33 , 85 – 89 , 92 , 93 ), 53.33% of studies reported the use of model fine-tuning ( 28 , 30 , 32 , 85 – 87 , 91 , 93 , 94 ), 73.33% of studies reported the use of internal validation methods ( 28 – 32 , 85 , 88 , 89 , 92 – 94 ), 13.33% of studies reported the use of external validation methods ( 28 , 30 ), 26.67% of studies reported that the data or the code used in the study can be made available upon reasonable request ( 28 , 30 , 33 , 92 , 93 ), 20% of studies included a link to the repository for their ML algorithm code ( 29 , 88 , 93 ), and 6.67% of studies included a link to the repository for the dataset used ( 87 ).…”

“…For treatment engagement and adherence prediction [n=2 ( 87 , 94 )], Kim et al. (2021) determined that higher engagement in an eCBT program for obesity was associated with higher weight loss.…”

“…This mixed-methods review aimed to present a comprehensive assessment of the current uses of AI in online mental healthcare using meta-analytic and network meta-analytic approaches. Our systematic search strategy identified various applications including triage, psychotherapy delivery, therapy progress monitoring, therapy engagement support, identification of engagement subtypes, effective therapy features and themes in patients’ utterances, and prediction of short-term and long-term treatment response, treatment dropout and adherence, and symptom remission ( 28 – 33 , 41 , 73 – 94 ). Analyzing these applications, the results of this review suggested that despite the novelty of this technology, AI has already demonstrated promising benefits to the field of online mental healthcare.…”