How Life Sciences is Leveraging Machine Learning - 5 Use Cases in 5 Minutes

The Panalgo “5 in 5” series outlines five current use cases where machine learning (ML) is being leveraged to analyze real-world data (RWD). Each edition of this series can be read in about five minutes, providing you with current use cases of   how life sciences and other healthcare entities are effectively utilizing ML in their research.

In this edition of “5 in 5,” Mike Munsell, Director of Research, highlights a range of current ML use cases – identifying drivers of prescribing patterns, predicting adverse outcomes, identifying at-risk patients for worse outcomes, and more. All different use cases, but all leveraging the power of ML to make predictions. 

Read on to see this issue's five.

1. Using Machine Learning to Examine Drivers of Inappropriate Outpatient Antibiotic Prescribing in Acute Respiratory Illnesses (The Centers for Disease Control and Prevention & IQVIA)

In a recent study published in Infection Control and Hospital Epidemiology, researchers from the CDC and IQVIA demonstrated the ability of machine learning to identify the drivers of inappropriate antibiotic prescribing. Evaluating these drivers helped the CDC target policy intervention. Read the full study here.

2. The Development and Validation of Simplified Machine Learning Algorithms to Predict Prognosis of Hospitalized Patients With COVID-19: Multicenter, Retrospective Study (Amgen Inc, Center for Observational Research & Amgen Inc, Digital Health & Innovation)

Researchers from Amgen recently published a study in The Journal of Medical Internet Research, where they used machine learning to develop a model for the prediction of mortality, intensive care unit admission, respiratory failure, or ventilator usage among hospitalized COVID-19 patients. Identifying and developing this simple 10-covariate risk algorithm for predicting adverse outcomes demonstrates the ability of machine learning to generate an interpretable risk score algorithm using high dimensional data. Read the full study here.

3. Identifying and Evaluating Clinical Subtypes of Alzheimer's Disease in Care Electronic Health Records Using Unsupervised Machine Learning (University College London, Health Data Research UK, Alan Turing Institute, & Amsterdam University Medical Centers)

For this study, researchers used unsupervised machine learning clustering techniques and real-world data to identify a meaningful subgroup of Alzheimer’s patients associated with worse clinical outcomes. They found that a subset of female patients with younger disease onset and comorbid depression and anxiety are more likely to have worse clinical outcomes and recommend further investigation of the health outcomes for this patient subtype. Read the full study published in BMC Medical Informatics and Decision Making  here.

4. Improving Patient Flow During Infectious Disease Outbreaks Using Machine Learning for Real-Time Prediction of Patient Readiness for Discharge (University of Oxford & Oxford University Hospitals NHS Foundation Trust)

In response to a shortage of hospital beds during the COVID-19 pandemic, Oxford researchers developed and validated a machine learning-based modeling technique to identify the patients with the highest likelihood of discharge readiness. The research published in PLOS ONE,  identified a model that may provide support and insights to clinicians for making hospital discharge decisions during periods of public health crises when resources are limited. Read the full study here.

5. Early Prediction of Hemodynamic Interventions in the Intensive Care Unit Using Machine Learning (Massachusetts Institute of Technology & Phillips Research North America)

In a recently published study in BMC, researchers developed a machine-learning based index for identifying hemodynamic instability in critically ill patients, rather than relying on basic vital signs such as heart rate, blood pressure, and shock index. The index may help clinicians catch instability and deliver intervention earlier. Read the full study here.

To learn more about Panalgo's IHD Data Science module, contact us.