Advanced Risk Assessment Methods Described in New Study

News provided by

Oct 20, 2022, 10:00 ET

Health Data Analytics Institute's Validated Approach May Help Guide Patient Management

BOSTON, Oct. 20, 2022 /PRNewswire/ -- Improving assessment of patient risk at hospital admission guides patient care and facilitates epidemiologic analyses. New research conducted by Health Data Analytics Institute (HDAI), in collaboration with researchers at the Cleveland Clinic, is being presented as a featured abstract at the 2022 Annual meeting of the American Society of Anesthesiologists on October 22 in New Orleans and will soon be published as a feature article in Anesthesiology, the Society's journal. The data show that predictive modeling based on administrative claims history provides individualized patient risk profiles at hospital admission that can guide patient management. The report is titled "Risk Stratification Index 3.0, a Broad Set of Models for Predicting Adverse Events During and After Hospital Admission."

"Our goal was to deliver a suite of easily deployable models for use by hospitals and health systems. Specifically, we developed and validated a broad suite of predictive tools based only on information available to a clinician before or on the day of admission to the hospital. Using individual medical claims history and demographic information, we accurately predicted risk for many adverse events up to ninety days post admission. We also observed similar outcomes from six different popular modeling approaches, which suggests that the predictive information derived from medical claims history can be extracted utilizing a variety of modeling techniques," said Nassib Chamoun, Founder, President, and CEO of HDAI.

Methodology: Predicted outcomes included excess length of stay, discharge status, unplanned hospital admissions, in-hospital and 90-day mortality, acute kidney injury, sepsis, pneumonia, respiratory failure, and a composite of major cardiac complications. Patient demographic and coding history in the year before admission provided features used to predict utilization and adverse events through 90-days post-admission. Models were trained and refined on 2017-18 Medicare admissions data using an 80-20 learn/test split sample. Models were then prospectively tested on 2019 out-of-sample Medicare admissions. Predictions based on logistic regression were compared with those from five commonly used machine learning methods using a separate limited dataset.

Results: The 2017-18 development set included 9,085,968 patients who had 18,899,224 inpatient admissions, and there were 5,336,265 patients who had 9,205,835 inpatient admissions in the 2019 validation dataset. Model performances on the validation set all exceeded the pre-specified acceptance criteria with an average area under the curve of 0.76 (Range 0.70, 0.82). Model calibration was very strong. Predictive accuracies from regression and machine learning techniques were generally similar.

"We are starting to deploy these tools for preoperative optimization and for developing risk-specific care paths to avoid complications." said Kamal Maheshwari, MD, MPH, Departments of General Anesthesiology and Outcomes Research at the Cleveland Clinic and one of the study authors. "Including a nationwide sample of millions of patients increases our confidence in practical application of these models to our work with patients before, during, and after surgery".

Health systems and Accountable Care Organizations use HDAI's predictive analytic product, named Health Vision, to support their care teams at the population and patient level, generating over 20M predictions every week. These innovative organizations are deploying the system as a stand-alone system fed by claims data in as little as 48 hours. Some are also enhancing the immediacy of the data through real-time EHR integration.

The Health Vision platform leverages the underlying models validated in this study, and enables care teams to rapidly identify patients at high risk for adverse events, along with the underlying factors that contribute to each risk. Subsequently, clinicians can leverage the patient's curated health history, including all encounters, medications, providers, tests, diagnoses, and interventions, to generate a personalized care plan, pre- and post-hospitalization.

The complete findings of the study, funded by HDAI, can be viewed online here and will be available in print in the coming months.

About HDAI

Health Data Analytics Institute (HDAI) is an analytics company with a versatile analytic platform that creates a shared understanding of quantified health risks and personalized care profiles to inform actions with the greatest potential to benefit patients. The predictors are built on extensive underlying data assets, driven by a sophisticated risk modeling methodology, and validated in multiple peer-reviewed articles. For more information, please visit: www.hda-institute.com.

Company contact: Carola Endicott, [email protected], 617-314-6135
Press contact: Erik Milster, [email protected]

SOURCE Health Data Analytics Institute