New Way to Use Data: Precision Care for Better Outcomes in Mental Health Care for Children and Teens

Program Description

Over 20% of children in our country suffer from mental health problems and only 20% of those children actually receive the help they need. When a child and family make the often hard decision to seek mental health services, providers need to be prepared to provide the right array of services to the child. Getting children the right treatment at the right time increases the likelihood of long term success. This project seeks to use data in an innovative way by utilizing machine learning to improve our clinical decision making so we can deliver the right treatment to the right child at the right time. Most providers collect data but it often only shows what has happened in the past. This project will enable us to use data to more precisely predict and prescribe what treatment interventions will be most effective with each child. The objectives include developing a learning community so together providers can develop the best algorithm that improves clinical decision making; embed the algorithm in our and other providers’ treatment process; and facilitate systemic change so that together we can improve the delivery of psychiatric residential treatment for children. *Please see attached (draft) paper detailing the project.

Creativity

To optimize success in the delivery of mental health services, providers need to deliver the right treatment to the right child at the right time. Traditionally, organizations collect data and report on what happened in the past. Though informative, this approach does not enable providers to determine what intervention is most effective for this child at this time. Our new approach to data uses machine learning (think Amazon) to more precisely predict and prescribe the treatment that has the most likely pathway to success for each child. This innovative use of data can revolutionize the way mental health treatment is delivered by improving clinical decision making and thereby our ability to match the right interventions with the right child. We know that getting children the right treatment at the right time vastly improves outcomes for children and teenagers and this project gives us the tools to make this a reality.

Leadership

Kristen Gay, Ph.D., President/CEO of Silver Springs has provided leadership and commitment to spearhead this project, championing its purpose, design, and implementation. Other organizations could follow a similar process by utilizing machine learning to analyze their data, model toward a desired outcome, and create an algorithm they can embed in their treatment process. Dr. Gay, along with our collaborating Data Scientist, has participated in meetings with foundation leaders, thought leaders in the City of Philadelphia, and CEOs/Executive Directors throughout the area. In addition she has sought interest at the national level through Lutheran Services in America in an effort to seek providers in other states to join our learning community. Dr. Gay and Silver Springs are using meetings, publications, and webinars to disseminate information about this project to educate and inspire others to partner with us as we pursue system-wide changes to the delivery of mental health care to children/adolescents.

Sustainability

Silver Springs is committed to embedding this innovative tool into its clinical decision making and has largely used its own financial resources combined with a small grant to complete phase one. Phase two includes partnering with other providers to learn from each other what data (set of variables) are most impactful in the precision care modeling. At least two other psychiatric residential treatment providers are interested in participating in the project. Through a shared learning community, we can compare what variables best predict future success and together improve our respective algorithms. A more precise algorithm will improve our ability to match the right interventions with the right child. Then our data scientist will design applications to embed the algorithm into our database. The ongoing expense of this technology will be minimal because these decision making tools will be available through our own applications available to front line clinicians.

Replicability

A major goal of this project is to create a shared learning platform so providers can learn from each other and together improve the delivery of psychiatric residential treatment for children. This concept has been of interest to other providers because many of us have been collecting data for years with variable results. We have gotten good at using our data to show how we performed historically, but we do not yet have the tools to use our data to predict and prescribe treatment. If we are to understand best what types of data predict future success, we need to facilitate the replication of this project so we can learn from each other. Part of the success of our project will depend on replicating phase one with other providers. The approach that Silver Springs is undertaking with Community Science’s partnership can easily be replicated and shared with other organizations.

Results/Outcomes

The outcome of this project will be to 1) create a learning platform for providers to learn from each other; 2) develop an algorithm to be used by front line clinicians to determine what interventions will best meet a child’s needs; and 3) eventually develop an algorithm for systemic level use. Getting children the right mental health treatment at the right time improves long term outcomes. We cannot afford to waste time and money delivering services that do not match the needs of the children being referred to our care. This innovative project will give us a tool to improve our ability to deliver the right treatment to the child at the right time, thereby increasing the utilization rates of mental health services because children and families will be delivered services that are helping, and also reducing the cost of service by decreasing the delivery of unnecessary/unhelpful services.