October 2, 2020 | Written by: Guillermo Cecchi
Categorized: AI | Healthcare
Share this post:
In collaboration with researchers from Harvard Medical School, Mt. Sinai School of Medicine, Stanford University and the Northern California Institute for Research and Education, IBM Research is undertaking a new research initiative funded by the National Institutes of Health.
As part of a broader $99 million, 5-year research initiative spanning multiple public and private organizations and research institutions, this work will tap into AI and big data to help better identify individuals at high-risk of developing schizophrenia, a serious mental illness affecting how a person thinks, feels and behaves. Schizophrenia is often characterized by alterations to a person’s thoughts, feelings and behaviors, which can include a loss of contact with reality known as psychosis. A better understanding of how this disease could be detected prior to psychosis could help to postpone or even prevent the transition to psychosis, as well as possibly improve outcomes. (1)
The project is a component of the Accelerating Medicines Partnership (AMP), a collaboration between the National Institutes of Health (NIH), the U.S. Food and Drug Administration (FDA), pharmaceutical companies, biotech firms and nonprofit organizations. Ultimately, projects launched by the AMP aim to increase the number of new diagnostics and therapies for patients, as well as reduce the time and cost of developing them.
For each project, scientists from the NIH and industry researchers seek to characterize effective molecular indicators of a disease, called biomarkers, and distinguish biological targets that are most likely to respond to new therapies. This same strategy will drive the research team’s work as they seek to unravel complexities and mysteries associated with schizophrenia and aim to understand better how it progresses and manifests within individuals.
While technological advances have enabled a wealth of data on the biological causes of disease, moving these discoveries into treatments is still incredibly difficult. Unfortunately, not all biological insights lead to effective drug targets. Choosing the wrong target can result in failures late in the drug development process, and cost enormous amounts of time, money and potentially even lives.
Developing a new drug, from early discovery to FDA approach, can take over a decade and has a failure rate of more than 95 percent. Often, the most expensive failures happen in late phase clinical trials, with a lack of drug efficacy currently estimated as responsible for 59 percent of Phase II failures, and 52 percent of Phase III failures. In the midst of a global pandemic, it’s especially apparent this current process needs to be accelerated, transformed and incredibly more efficient. Pinpointing viable biological targets early in the drug development process can help to catalyze this change. (2)
The entire biomedical research community and the public have a shared interest in compressing the timelines, reducing the costs, and increasing the success rates of new targeted therapies. Given the amount and complexity of the data, this goal will require a systematic approach in which government, academia, industry, and patient groups work collaboratively to sift through the flood of disease targets and find the ones most likely to prove responsive to treatments.
By optimizing the process for identifying and validating clinically relevant disease targets for drug design, the projects undertaken by the AMP aim to shorten development time, improve prospects of success, lower costs and number of Phase II and III failures, and gain understanding of biological targets, pathways and biomarkers.
As the next rollout of AMP, NIH has launched a Schizophrenia initiative (AMP SZ), with the participation of the pharmaceutical industry, including Boehringer Ingelheim, Janssen, Otsuka and Raritan, and foundations including NAMI (National Alliance of Mental Health), the American Psychiatric Association (APA) Foundation, and the Wellcome Foundation. In addition, the initiative will coordinate its activities with the Federal Drug Administration (FDA). AMP SZ will be funded for 5 years with a total budget of 99 million to support efforts in data collection and data analytics. (3)
The AMP SZ initiative will consist of two centers for data collection and one center for data analysis, including the Clinical High Risk for Psychosis: Data Processing, Analysis, and Coordination Center (DPACC). The consortium known as Psychosis Risk Evaluation, Data Integration and Computational Technologies (PREDICT) will consist of researchers from Harvard Medical School, Mt. Sinai School of Medicine, Stanford University, the Northern California Institute for Research and Education, and IBM Research. These teams were chosen by NIMH to be the DPACC Center.
To be undertaken across five years, our work aims to help address the heterogeneity of those at clinical high risk for developing psychosis, where such variability in outcome has impeded progress in treatment development. The goals of the PREDICT Center are to bring together a multidisciplinary team of highly experienced researchers with proven capabilities in all aspects of large-scale studies, as well as computational expertise, to develop biomarker algorithms, first in aggregate extant data sets, and then in prospective new data collected by the research network investigators who are partners in the Schizophrenia AMP.
The aim is for these biomarkers to be used to help predict individual clinical high risk trajectories and individual outcomes, and to generate individualized risk calculators that can be used in future clinical trials for treatment intervention and, hopefully, to halt the progression of adverse outcomes that include psychosis onset, and decline in functioning, as well as spectrum of other outcomes ranging from persistent attenuated symptoms, to anxiety and mood/disorders, to alcohol and drug abuse, to suicidal ideation and behavior, to recovery.
The “Clinical High Risk” (CHR) for psychosis syndrome in young people represents an opportune window for early intervention to help prevent the onset of psychosis and other disorders, and to forestall disability; however, clinical heterogeneity and the paucity of biomarkers have hampered the development of effective intervention.
To address these challenges, working with NIMH and key stakeholders, this initiative will strive to harmonize and aggregate existing “legacy” CHR data, as well as guide clinical partners as they collect new data across a network of sites. Working with this database of information aggregated by separate data collection centers, our team will work to develop biomarker algorithms that can help predict individual trajectories for diverse outcomes. This proposal leverages a multidisciplinary team with broad and CHR-specific experience in large-scale multisite and multimodal studies (including clinical trials), along with expertise in data type-specific processing, coordination, analysis, and computational analyses, ethics, community outreach, and data dissemination, all of which will ensure the success of this project.
The IBM Research team will contribute its knowledge in data-driven artificial intelligence applications to brain imaging for neurodevelopmental and neurodegenerative disorders, as demonstrated in schizophrenia, chronic pain and Huntington’s disease. The team will also contribute its knowledge to analyze and guide the collection of language samples, based on the track record of successful application of Natural Language Processing approaches to predict onset of psychosis in the CHR population. Dr. Guillermo Cecchi will be an Associate Director on this project and co-leader of the Data Analytics Core, coordinating the efforts of the IBM team. He will be working closely with Dr. Polosecki, Castro and Reinen, also on the IBM Research team, across the other cores of the project, including data management, data monitoring and dissemination.
The PREDICT Center and the initiative as a whole will be a unique opportunity for IBM Research to be a leader in the realization the enormous potential of integrating large volumes of data, artificial intelligence and basic neuroscientific research to help impact mental health.