A machine learning expert and a psychology researcher/clinician might seem like an unlikely duo. But the MIT Rosalind Picard and Massachusetts General Hospital paola petrelli they are united by the belief that artificial intelligence can help make mental health care more accessible to patients.
In her 15 years as a clinician and researcher in psychology, Pedrelli says that “it has been very, very clear that there are a number of barriers for patients with mental health disorders to access and receive appropriate care.” Those barriers can include figuring out when and where to seek help, finding a nearby provider who accepts patients, and obtaining financial resources and transportation to get to appointments.
Pedrelli is assistant professor of psychology at Harvard Medical School and associate director of the Depression Clinical and Research Program at Massachusetts General Hospital (MGH). For more than five years, he has been collaborating with Picard, a professor of media arts and sciences at MIT and principal investigator of the Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic) on a project to develop machine learning algorithms to help diagnose and monitor changes in symptoms among patients with major depressive disorder.
Machine learning is a type of artificial intelligence technology where, when the machine is given a lot of data and examples of good behavior (i.e. what output to produce when it sees a particular input), it can become quite good at performing a task autonomously. It can also help identify patterns that are meaningful, that humans might not have been able to find as quickly without the help of the machine. Using wearable devices and smartphones from study participants, Picard and Pedrelli are able to collect detailed data on participants’ skin conductance and temperature, heart rate, activity levels, socialization, self-assessment of depression , sleep patterns and more. Their goal is to develop machine learning algorithms that can absorb this huge amount of data and make it meaningful, identifying when a person might be struggling and what might be useful to them. They hope their algorithms will eventually provide doctors and patients with useful information about individual disease trajectories and effective treatment.
“We are trying to build sophisticated models that have the ability not only to learn what is common between people, but also to learn categories of what is changing in an individual’s life,” says Picard. “We want to give people who want it the opportunity to have access to personalized, evidence-based information that makes a difference to their health.”
Machine learning and mental health
Picard joined the MIT Media Lab in 1991. Three years later, he published a book, “Affective Computing,” which prompted the development of a field by that name. Affective computing is now a strong area of research related to the development of technologies that can measure, sense, and model data related to people’s emotions.
While early research focused on determining whether machine learning could use data to identify a participant’s current emotion, the current work by Picard and Pedrelli at MIT’s Jameel Clinic goes several steps further. They want to know if machine learning can estimate the trajectory of the disorder, identify changes in an individual’s behavior, and provide data that informs personalized healthcare.
Picard and Szymon Fedor, a research scientist in Picard’s affective computing lab, began collaborating with Pedrelli in 2016. After conducting a small pilot study, they are now in the fourth year of their five-year National Institutes-funded study. of health.
To conduct the study, the researchers recruited MGH participants with major depressive disorder who had recently changed their treatment. So far, 48 participants have been enrolled in the study. For 22 hours a day, every day for 12 weeks, participants wear Empatica E4 bracelets. These wearable wristbands, designed by one of the companies Picard founded, can collect information about biometric data, such as electrodermal (skin) activity. Participants also download apps onto their phones that collect data on text messages and phone calls, location, and app usage, and also ask them to complete a biweekly depression survey.
Each week, patients see a doctor who assesses their depressive symptoms.
“We put all the data we collect from the wearable and the smartphone into our machine learning algorithm, and we try to see how well the machine learning predicts the labels given by doctors,” says Picard. “Right now, we’re pretty good at predicting those labels.”
Empowering users
While developing effective machine learning algorithms is one challenge facing researchers, designing a tool that empowers and uplifts its users is another. Picard says, “The question we’re really focusing on now is, once you have machine learning algorithms, how is that going to help people?”
Picard and his team are thinking critically about how machine learning algorithms can present their findings to users: through a new device, a smartphone app, or even a method of notifying a doctor or predetermined family member about the disease. best way to help the user.
For example, imagine technology that records that a person has recently been sleeping less, spending more time indoors, and has a faster-than-usual heart rate. These changes can be so subtle that the person and their loved ones have not yet noticed them. Machine learning algorithms can make sense of this data, mapping it to the individual’s past experiences and the experiences of other users. The technology can then encourage the person to engage in certain behaviors that have improved their well-being in the past, or to communicate with their doctor.
If implemented incorrectly, it is possible that this type of technology could have adverse effects. If an app alerts someone that they are headed for deep depression, it could be discouraging information that leads to more negative emotions. Pedrelli and Picard are involving real users in the design process to create a tool that is helpful, not harmful.
“What might be effective is a tool that says to an individual, ‘The reason you’re feeling depressed might be that the data related to your sleep has changed, and the data is related to your social activity, and you haven’t had time with your friends, your physical activity has decreased. The recommendation is that you find a way to increase those things,'” Picard says. The team is also prioritizing data privacy and informed consent.
Artificial intelligence and machine learning algorithms can make connections and identify patterns in large data sets that humans aren’t as good at noticing, says Picard. “I think there’s a really compelling case for technology to help people be smarter with people.”
.
