Access to patient-specific data—by patients and healthcare organizations alike—is key to improving health outcomes, but with 75% of clinical data found in unstructured form and more than 100 million clinical record reviews performed each year in the United States, it can be challenging to get the data you need for a particular patient at a given time. Addressing this challenge head on, Inovalon’s natural language processing solution accelerates and automates the clinical record review process, delivering rapid insight into clinical quality, risk, disease outcomes and costs.
Inovalon’s Natural Language Processing as a Service (NLPaaS™) leverages the Inovalon ONE® Platform’s highly adaptive machine learning capabilities along with data on billions of known medical events to transform unstructured clinical data into highly valuable, structured data, which are then analyzed at massive scale in fractions of a second. Employing state-of-the-art deep learning and machine learning algorithms, our natural language processing solution significantly decreases review time, enabling risk score-relevant condition detection 4 times faster than traditional methods. Our natural language processing solution can process up to 35,000 records per day, reducing the overall clinical review program spend by up to 25% for our clients.
Inovalon’s Risk Gap Detection solution uses categorical claims data such as NDC, ICD, or CPT codes to identify gaps in care not indicated in the patient’s record. Leveraging machine learning to supplement this process and integrating thousands of additional variables (including procedure, pharmaceutical and lab data), Inovalon’s solution identifies with high confidence the probability of a gap being present for a particular patient. More than two decades of clinical review and regulatory subject matter expertise empower the Risk Gap Detection solution, which can be retrained and deployed upon any change to risk score accuracy models, empowering our clients to access real-time insights and drive greater risk score accuracy.
Predicting patient-specific events is a daunting task that requires an immense volume and depth of data. Leveraging a time-aware long short-term neural network (T-LSTM), Inovalon’s Future Encounters Prediction Framework analyzes more than 70,000 ICD codes and more than 100,000 claims code data to determine in real-time the predicted outcome of a particular patient’s adverse event.
With successful applications including predictions of fall risk, stroke, and emergency admissions, the Framework is easily extendable to any future encounter prediction. The solution provides performance data, as well as transparency, as it not only forecasts an event but also provides transparency into the computer’s decisions. It is one of the only models of its kind today, made possible through Inovalon’s proprietary MORE2 Registry®.
Population health analytics are important for both providers and payers to help manage the morbidity of specific patients as well as the overall health of a patient population. The ability to stratify risk and predict costs, utilization, medication adherence and readmissions in real-time allows for a data-driven approach to care coordination and management.
Leveraging artificial intelligence, Inovalon’s population health analytics provide real-time insights into patients and their risk level relative to broader populations, allowing providers and payers to focus on key population health outcomes. Inovalon’s solution integrates a variety of data streams, including social determinants of health, claims, clinical, and census data to help clients more accurately design a care program that is population centric, with the ability to stratify by specific quality measurement outcomes.
Predicting the number of patients that are expected to be in a hospital during a given time period—the “census”—is extremely important for determining hospital or department-specific staffing needs for the next day, days, or weeks. But predicting hospital census is challenging, as data available often varies significantly from practice to practice and a single, scalable solution can be difficult to develop and employ.
To address this issue, Inovalon developed a Hospital Census Prediction solution, utilizing the MORE2 Registry®. The solution forecasts patient load based on time series analysis with minimal information, providing insight into what future value will be, as well as future distribution of that value. Providers can use the solution to dramatically decrease overall operational costs, improving the efficiency of the workforce.