Decoding Population Health Analytics: Powering Pharma R&D and Clinical Decision Support

Advanced analytics are imperative to improve patient outcomes and control costs in today’s healthcare system. These methods are useful for pharmaceutical companies to refine inclusion/exclusion criteria for clinical trials and for provider organizations to design care management programs that ensure optimal health outcomes and control high-cost events. However, challenges abound to unlock value. Building an effective data program can be challenging, whether it involves cleaning and structuring large datasets or navigating technical and technological choices. Decode Health’s AI/ML platform makes advanced analytics exploration easy. We help healthcare decision-makers predict population health outcomes and identify individuals at risk for specific health issues, including high-cost events. The Decode platform also offers valuable insights into clinical and social determinants of health (SDoH) risk factors linked to health outcomes. This technology improves patient outcomes, optimizes interventions, and streamlines clinical research and trials.

Our platform utilizes advanced algorithms to detect specific data patterns, enabling accurate predictions. We assist R&D teams and healthcare professionals in making informed decisions faster. This is critical in improving the efficiency of clinical trials, managing acute or chronic conditions, and responding to public health crises efficiently. The Decode team works directly with existing customer data but has established an ecosystem of technology partners and data providers to quickly test and commercialize new solutions. AI/ML solutions should be tailored to fit a population’s unique characteristics and account for subtle differences in populations from different communities. Decode’s partner ecosystem is also leveraged to create new data assets, especially for multiomics studies lacking readily available datasets.

Further, our data quality control processes ensure the reliability of results. These processes examine the reproducibility of specific data inputs and the sustainability of those input data sources to ensure that deployed data models generate commercial value. We prioritize using high-quality data inputs to provide accurate AI/ML modeling results. It is essential to address issues related to data quality in real-world healthcare settings.

Use Case: Oncology Patient Monitoring 

Decode recently collaborated with a global pharmaceutical company to analyze healthcare claims data and SDoH associated with diffuse large B cell lymphoma (DLBCL) and non-small cell lung cancer (NSCLC) to understand high-risk patient personas, including cancer patients who were at the greatest risk of poor financial outcomes compared to their disease peers. After cleaning and processing data from three distinct states in the U.S., our platform built and selected models to predict high-risk, high-cost DLBCL and NSCLC patients. The resulting models predicted 90% of the total spend for high-cost DLBCL patients over four months, compared to conventional methods capturing only 42% of the total spend. Tracking specific high-risk individuals and understanding the risk factors leading to worse outcomes can influence care plans and the design of clinical trials to find patients who can benefit the most from treatment, leading to more efficient use of resources.

Use Case: Autoimmune Disease Detection and Monitoring

Another collaboration with a global technology firm further highlights the platform’s predictive power. In this second claims analysis project, Decode’s predictive approach was tested in a simulated environment to detect Crohn’s disease and monitor a group of Crohn’s patients to anticipate high-cost events. Representative patient journeys included one patient initially diagnosed with ulcerative colitis that Decode identified as a patient with Crohn’s. Longitudinal follow-up of the patient confirmed a subsequent, corrected diagnosis of Crohn’s. Compared to other Crohn’s patients, this patient’s cost profile was 3X higher, mainly due to inadequate treatment. Similarly, while monitoring Crohn’s patients for high-cost events, one high-risk Crohn’s patient was identified four months before a significant hospitalization event, resulting in a cost profile 10X higher than the average healthcare spending profile of others in the population. These insights indicate how advanced analytics by Decode can lead to earlier interventions, improving outcomes for patients with complex chronic conditions.

Decode Health’s AI/ML data platform is disease-agnostic and integrates a wide range of healthcare data sources, including electronic health records (EHR), insurance claims, SDoH information, patient survey data, and more, to detect, monitor, and predict specific health outcomes. The platform can zoom in to pinpoint specific members of a population and zoom out to spotlight geographic and population-level trends that are associated with disease outcomes. This approach enables targeted interventions and optimizes clinical research, trials, and care management. The adaptable approach reduces inefficiencies common in precision medicine solution development.

To learn more, email Decode Health at info@decodehealth.ai