Patient-Centered Pain Care Using Artificial Intelligence and Mobile Health Tools: Protocol for a Randomized Study Funded by the US Department of Veterans Affairs Health Services Research and Development Program

Abstract

BackgroundCognitive behavioral therapy (CBT) is one of the most effective treatments for chronic low back pain. However, only half of Department of Veterans Affairs (VA) patients have access to trained CBT therapists, and program expansion is costly. CBT typically consists of 10 weekly hour-long sessions. However, some patients improve after the first few sessions while others need more extensive contact.ObjectiveWe are applying principles from “reinforcement learning” (a field of artificial intelligence or AI) to develop an evidence-based, personalized CBT pain management service that automatically adapts to each patient’s unique and changing needs (AI-CBT). AI-CBT uses feedback from patients about their progress in pain-related functioning measured daily via pedometer step counts to automatically personalize the intensity and type of patient support. The specific aims of the study are to (1) demonstrate that AI-CBT has pain-related outcomes equivalent to standard telephone CBT, (2) document that AI-CBT achieves these outcomes with more efficient use of clinician resources, and (3) demonstrate the intervention’s impact on proximal outcomes associated with treatment response, including program engagement, pain management skill acquisition, and patients’ likelihood of dropout.MethodsIn total, 320 patients with chronic low back pain will be recruited from 2 VA healthcare systems and randomized to a standard 10 sessions of telephone CBT versus AI-CBT. All patients will begin with weekly hour-long telephone counseling, but for patients in the AI-CBT group, those who demonstrate a significant treatment response will be stepped down through less resource-intensive alternatives including: (1) 15-minute contacts with a therapist, and (2) CBT clinician feedback provided via interactive voice response calls (IVR). The AI engine will learn what works best in terms of patients’ personally tailored treatment plans based on daily feedback via IVR about their pedometer-measured step counts, CBT skill practice, and physical functioning. Outcomes will be measured at 3 and 6 months post recruitment and will include pain-related interference, treatment satisfaction, and treatment dropout. Our primary hypothesis is that AI-CBT will result in pain-related functional outcomes that are at least as good as the standard approach, and that by scaling back the intensity of contact that is not associated with additional gains in pain control, the AI-CBT approach will be significantly less costly in terms of therapy time.ResultsThe trial is currently in the start-up phase. Patient enrollment will begin in the fall of 2016 and results of the trial will be available in the winter of 2019.ConclusionsThis study will evaluate an intervention that increases patients’ access to effective CBT pain management services while allowing health systems to maximize program expansion given constrained resources.

General Practitioners’ Perspective on eHealth and Lifestyle Change: Qualitative Interview Study

“…Studies show that PROM data could be supported by algorithms based on eHealth [ 29 ] and through patient-centered care [ 30 , 31 ], giving space for the patient to act. Adding machine learning or artificial intelligence to PROM data could potentially strengthen the GPs’ ability to assist patients cost efficiently in conjunction with other health care professionals, which is being tested in specific patient groups [ 32 ], but will need substantially more research.…”

Internet‐Based Pain Self‐Management for Veterans: Feasibility and Preliminary Efficacy of the Pain EASE Program

“…Given that participants logged onto the website an average of 17 minutes at each use, it is possible that they accessed more than 1 (or several) modules at each logon, as they were not restricted to certain modules or a certain number of modules in any given week of the program. In addition, the average number of skill modules completed is consistent with a “dose” of CBT, (ie, 3 sessions, as defined in other trials by this group using the same CBT content) …”

“…In addition, the average number of skill modules completed is consistent with a "dose" of CBT, (ie, 3 sessions, as defined in other trials by this group using the same CBT content). 18,19,34,35 Engaging participants in the program is an important task. It is likely beneficial that the Pain EASE program is device agnostic (ie, as readable and usable on a mobile device as it is on a computer), providing flexibility for using the program without requiring a desktop or laptop computer.…”

Toward Comprehensive Patient-Centric Care by Integrating Digital Health Technology With Direct Clinical Contact in Australia

“…Symptom monitoring and patient feedback through the BHS could be used to assign patients to the appropriate level of stepped care, and the provision of care tailored to the individual’s level of need could be delivered using Web-based interventions through the system. Piette et al [57] have developed a personalized cognitive behavior therapy pain management service that adapts to each patient’s unique and changing needs to automatically personalize the intensity and type of patient support, using feedback from patients about their progress. Technology would enable the tailoring to individual, clinical, and personal circumstances by using reinforcement learning algorithms to deliver information that is most relevant to each user, similar to those algorithms used by Netflix, Google, and Amazon [58].…”