Skip to main content
ABCT 2024 58th Annual Convention Main banner
Psychology Tools Ad
Final Program


Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Presentation Icons
Recorded Session
Recorded Session
Ticketed Session
Ticketed Session
Back
ViewAttendees 4

Symposia

Technology/Digital Health

The promise of technological advancements in innovating various aspects of healthcare delivery

5 - (SYM 134) Development of Machine Learning Models to Identify and Match Treatment Recommendations

Friday, November 15, 2024
8:30 AM – 10:00 AM EST
Location: Congress B, Fourth Floor, Loews Hotel
Abstract Body

In an era of rising mental health care needs worldwide, advancement in technologies can help address care needs in a cost-effective fashion. Precision Behavioral Health (PBH) is a care model that matches patients to evidence-based digital interventions with strategic human support within routine care. The present study describes the development of a patient-centered machine learning approach embedded within PBH that identifies distinctive clinical profiles within patients, and how these relate to different treatment options as well as corresponding treatment outcomes. Latent Profile Analysis, an application of machine learning (Kumar et al., 2010), was used to identify distinct patient clinical profiles using a multidimensional measure, Norse Feedback. For each of the patient profiles, we analyzed how different treatment referral options (short-term therapy, long-term therapy, digital interventions) corresponded with differential treatment outcomes. Data included 4964 patients that attended a triage assessment with a behavioral health integrated clinician between 2022-12-01 and 2023-12-31. We used an analytically derived approach to remove patients who had a profile with less than 0.75 posterior probability to reach a final sample size of 2090 patients. The results showed 33 distinct patient clinical profiles. For example, one patient profile that was predominantly characterized by anger symptoms had highest referral rates to short-term therapy, and more than 50% of referred patients showed clinically significant symptom improvement. Another example profile included patients with elevated social avoidance, was more likely to be referred to two digital interventions, and these interventions had higher percentage of patients with clinically significant improvement compared to other interventions. The results of the study show how machine learning based identification of patients could optimize treatment recommendations based on improved outcomes. These types of technological advances can have significant public health implications as they can be added to existing clinical workflows as part of a clinical decision support tool to aid provider care recommendations.