Summary: A new multitasking model artificial intelligence algorithm based on data from wearable devices predicts treatment outcomes individually for people with depression.
Font: WUSTL
In recent years, mental health management has become more of a priority with a greater emphasis on self-care. Depression alone affects more than 300 million people worldwide annually.
Recognizing this, there is great interest in leveraging popular wearable devices to monitor an individual’s mental health by measuring markers such as activity levels, sleep, and heart rate.
A team of researchers from Washington University in St. Louis and the University of Illinois at Chicago used data from wearable devices to predict depression treatment outcomes in people who participated in a randomized clinical trial.
They developed a new machine learning model that analyzes data from two sets of patients, those randomly selected to receive treatment and those who did not receive treatment, rather than developing a separate model for each group.
This unified multitasking model is a step toward personalized medicine, in which clinicians design a treatment plan specific to each patient’s needs and predict the outcome based on an individual’s data.
The results of the research were published in the ACM Proceedings on Interactive, Model, Wearable and Ubiquitous Technologies and will be presented at the UbiComp 2022 conference in September.
Chenyang Lu, a Fullgraf professor at the McKelvey School of Engineering, led a team that included Ruixuan Dai, who worked in Lu’s lab as a doctoral student and is now a software engineer at Google; Thomas Kannampallil, associate professor of anesthesiology and associate director of research information in the School of Medicine and associate professor of computer science and engineering at McKelvey Engineering; and Jun Ma, MD, PhD, professor of medicine at the University of Illinois Chicago (UIC); and colleagues to develop the model using data from a randomized clinical trial conducted by UIC with approximately 100 adults with depression and obesity.
“Integrated behavioral therapy can be expensive and time consuming,” Lu said.
“If we can make personalized predictions for people about whether a patient is likely to respond to a particular treatment, then patients can proceed with treatment only if the model predicts that their conditions are likely to improve with treatment, but less likely to improve. No treatment. Such personalized predictions of treatment response will facilitate more targeted and cost-effective therapy.”
In the trial, patients received Fitbit bracelets and psychological tests. About two thirds of the patients received behavioral therapy and the rest of the patients did not. Patients in both groups were statistically similar at the start of the study, giving the researchers a level playing field to discern whether the treatment would lead to better outcomes based on individual data.
Clinical trials of behavioral therapies often involved relatively small cohorts due to the cost and duration of such interventions. The small number of patients created a challenge for a machine learning model, which typically works better with more data.
However, by combining the data from the two groups, the model was able to learn from a larger data set, which captured differences between those who had undergone treatment and those who had not. They found that their multitasking model predicted depression outcomes better than a model that looked at each of the groups separately.
“We pioneered a multitasking framework, combining the intervention group and the control group in a randomized control trial to jointly train a unified model to predict an individual’s personalized outcomes with and without treatment,” said Dai, who obtained a doctorate in computing. science in 2022.
“The model integrated clinical features and portable data in a multilayer architecture. This approach avoids dividing study cohorts into smaller groups for machine learning models and enables dynamic knowledge transfer between groups to optimize prediction performance with and without intervention.”
“The implications of this data-driven approach extend beyond randomized clinical trials to implementation in clinical care delivery, where the ability to make personalized prediction of patient outcomes based on treatment received, and to do so early and throughout the course of treatment, it could significantly inform shared decision-making by the patient and the treating physician in order to tailor the treatment plan for that patient,” Ma said.
The machine learning approach provides a promising tool to build personalized predictive models based on data collected from randomized controlled trials.
In the future, the team plans to leverage the machine learning approach in a new randomized controlled trial of behavioral telehealth interventions using Fitbit wristbands and scales among patients in a weight loss intervention study.
See also
About this research news on neurotechnology and depression
Author: brandy jefferson
Font: WUSTL
Contact: Brandie Jefferson – WUSTL
Image: The image is in the public domain.
original research: Open access.
“Multitask learning for randomized controlled trials: a case study on predicting depression with wearable data.by Chenyang Lu et al. ACM Proceedings on Ubiquitous and Portable Mobile Interactive Technologies
Summary
Multitask learning for randomized controlled trials: a case study on predicting depression with wearable data.
A randomized controlled trial (RCT) is used to study the safety and efficacy of new treatments, comparing the results of patients in an intervention group with a control group. Traditionally, RCTs rely on statistical analyzes to assess differences between treatment and control groups.
However, such statistical analyzes are generally not designed to assess the impact of the intervention at the individual level. In this article, we explore machine learning models coupled with an RCT for personalized predictions of a depression treatment intervention, where patients were monitored longitudinally with wearable devices.
We make predictions at the individual level in the intervention and control groups of an RCT as a multitasking learning problem (MTL), and we propose a new MTL model designed specifically for RCTs. Instead of training separate models for the intervention and control groups, the proposed MTL model is trained on both groups, effectively expanding the training dataset.
We developed a hierarchical model architecture to aggregate data from different sources and different longitudinal stages of the trial, allowing the MTL model to exploit commonalities and capture differences between the two groups. We evaluated the MTL approach in an RCT involving 106 patients with depression, who were randomized to receive an integrated treatment intervention.
Our proposed MTL model outperforms both single-task and traditional multi-task models in predictive performance, representing a promising step in utilizing data collected in RCTs to develop predictive models for precision medicine.
