Personalized Depression Treatment

For a lot of people suffering from depression, traditional therapy and medication isn't effective. Personalized treatment could be the answer.

Cue is an intervention platform for digital devices that converts passively collected smartphone sensor data into personalized micro-interventions to improve mental health. We looked at the best-fitting personal ML models to each subject using Shapley values to discover their features and predictors. The results revealed distinct characteristics that changed mood in a predictable manner over time.

Predictors of Mood

Depression is the leading cause of mental illness across the world.1 Yet the majority of people suffering from the condition receive treatment. To improve outcomes, clinicians must be able to recognize and treat patients most likely to respond to certain treatments.

A customized depression treatment is one method to achieve this. Using sensors on mobile phones as well as an artificial intelligence voice assistant and other digital tools, researchers at the University of Illinois Chicago (UIC) are working on new ways to determine which patients will benefit from the treatments they receive. Two grants worth more than $10 million will be used to identify the biological and behavioral indicators of response.

So far, the majority of research into predictors of depression treatment effectiveness has focused on the sociodemographic and clinical aspects. These include demographic variables such as age, gender and education, clinical characteristics including symptom severity and comorbidities, and biological indicators such as neuroimaging and genetic variation.

Few studies have used longitudinal data to predict mood of individuals. A few studies also consider the fact that mood can be very different between individuals. Therefore, it is critical to develop methods that permit the identification of different mood predictors for each person and treatments effects.

The team's new approach uses daily, in-person evaluations of mood and lifestyle variables using a smartphone app called AWARE, a cognitive evaluation with the BiAffect app and electroencephalography -- an imaging technique that monitors brain activity. This allows the team to create algorithms that can identify various patterns of behavior and emotion that differ between individuals.

The team also created an algorithm for machine learning to model dynamic predictors for each person's depression mood. The algorithm blends these individual differences into a unique "digital phenotype" for each participant.

This digital phenotype was linked to CAT DI scores that are a psychometrically validated symptoms severity scale. However, the correlation was weak (Pearson's r = 0.08, the BH-adjusted p-value was 3.55 x 10-03) and varied widely among individuals.

Predictors of symptoms

Depression is one of the most prevalent causes of disability1 but is often untreated and not diagnosed. Depression disorders are rarely treated because of the stigma attached to them, as well as the lack of effective treatments.

To assist in individualized treatment, it is crucial to identify predictors of symptoms. However, the methods used to predict symptoms are based on the clinical interview, which has poor reliability and only detects a tiny number of features related to hormonal depression treatment.2

Machine learning can improve the accuracy of diagnosis and treatment for depression by combining continuous, digital behavioral patterns gathered from sensors on smartphones with a valid mental health tracker online (the Computerized Adaptive Testing Depression Inventory CAT-DI). These digital phenotypes allow continuous, high-resolution measurements and capture a wide variety of distinctive behaviors and activity patterns that are difficult to record using interviews.

The study included University of California Los Angeles students who had mild to severe depression symptoms who were participating in the Screening and Treatment for anxiety depression treatment (Continued) and Depression program29, which was developed as part of the UCLA Depression Grand Challenge. Participants were directed to online support or clinical care according to the degree of their alcohol depression treatment. Those with a score on the CAT DI of 35 or 65 were given online support with the help of a coach. Those with scores of 75 patients were referred for psychotherapy in person.

At baseline, participants provided an array of questions regarding their personal demographics and psychosocial characteristics. The questions covered age, sex, and education as well as financial status, marital status and whether they were divorced or not, their current suicidal thoughts, intent or attempts, and how often they drank. Participants also rated their level of depression severity on a 0-100 scale using the CAT-DI. The CAT-DI assessment was performed every two weeks for those who received online support and weekly for those who received in-person support.

Predictors of Treatment Response

Research is focusing on personalized treatment for depression. Many studies are focused on identifying predictors, which will help doctors determine the most effective medications for each person. Pharmacogenetics in particular is a method of identifying genetic variations that affect how the human body metabolizes drugs. This allows doctors to select drugs that are likely to be most effective for each patient, minimizing the time and effort in trial-and-error treatments and eliminating any side effects that could otherwise hinder the progress of the patient.

Another promising method is to construct models of prediction using a variety of data sources, combining clinical information and neural imaging data. These models can be used to identify the best combination of variables that are predictors of a specific outcome, such as whether or not a particular medication is likely to improve mood and symptoms. These models can be used to determine the patient's response to a treatment, allowing doctors to maximize the effectiveness.

A new generation of studies employs machine learning techniques such as supervised learning and classification algorithms (like regularized logistic regression or tree-based methods) to combine the effects of multiple variables and improve the accuracy of predictive. These models have proven to be useful for predicting treatment outcomes such as the response to antidepressants. These approaches are gaining popularity in psychiatry, and it is expected that they will become the norm for future clinical practice.

Research into depression's underlying mechanisms continues, in addition to predictive models based on ML. Recent findings suggest that depression is related to dysfunctions in specific neural networks. This theory suggests that an individualized treatment for depression treatment plan will depend on targeted treatments that restore normal function to these circuits.

One method of doing this is through internet-delivered interventions which can offer an personalized and customized experience for patients. For example, one study found that a program on the internet was more effective than standard treatment in improving symptoms and providing a better quality of life for people suffering from MDD. A controlled, randomized study of a customized treatment for depression showed that a significant percentage of patients saw improvement over time and fewer side consequences.

Predictors of adverse effects

A major challenge in personalized depression treatment involves identifying and predicting the antidepressant medications that will have minimal or no side effects. Many patients have a trial-and error approach, with a variety of medications prescribed until they find one that is safe and effective. Pharmacogenetics offers a fascinating new way to take an efficient and specific approach to selecting antidepressant treatments.

There are many predictors that can be used to determine the antidepressant that should be prescribed, such as gene variations, phenotypes of patients such as gender or ethnicity and co-morbidities. To determine the most reliable and reliable predictors of a specific treatment, randomized controlled trials with larger samples will be required. This is due to the fact that it can be more difficult to identify interactions or moderators in trials that comprise only a single episode per person instead of multiple episodes spread over a period of time.

Additionally to that, predicting a patient's reaction will likely require information on comorbidities, symptom profiles and the patient's personal perception of effectiveness and tolerability. Presently, only a handful of easily assessable sociodemographic and clinical variables appear to be reliable in predicting response to MDD like gender, age race/ethnicity, SES BMI, the presence of alexithymia, and the severity of depression symptoms.

There are many challenges to overcome in the use of pharmacogenetics for depression treatment. It is crucial to have a clear understanding and definition of the genetic factors that cause depression, as well as an understanding of an accurate indicator of the response to treatment. Additionally, ethical issues like privacy and the ethical use of personal genetic information, should be considered with care. The use of pharmacogenetics may eventually help reduce stigma around treatments for mental illness and improve the quality of treatment. As with all psychiatric approaches it is essential to carefully consider and implement the plan. In the moment, it's best to offer patients a variety of medications for depression that are effective and urge them to speak openly with their doctors.