Machine learning based diagnosis of heart failure with preserved ejection fraction among South Asian patients

Funding Acknowledgements

Type of funding sources: None.

Background

Diagnosis of Heart Failure with preserved Ejection Fraction (HFpEF) remains challenging even for an astute clinician with the current diagnostic tests, algorithms and scoring systems.  The use of Machine Learning (ML) in cardiovascular research has expanded exponentially in recent years.

Purpose

In this study we studied the predictive accuracy of a machine-learning model for the diagnosis of HFpEF in South Asian patients.

Methods

A single centre retrospective data of 1068 admitted patients with a clinical diagnosis of HF were obtained.  From this data outliers were removed based on box-and-whisker plot study.  This data was then segregated into three: without HF, HF with reduced ejection fraction and HFpEF.  Finally, we arrived at 530 patients’ data with left ventricular ejection fraction (LVEF) values of >45%. This data set was randomly divided into two parts: 80% as Training dataset and 20% as Test dataset.  Two ML techniques such as Logistic Regression and Neural Networks were chosen to build the prediction models. Orange Data Mining Software version 3.28 was used to build these models. The overall performance of the ML methods has been reported using Precision, Recall and F1 scores of precision and recall; measure that combines precision and recall.

Results

The table summarizes precision, recall and F1 score of both models. The predicted results from the ML models are compared with HFA PEFF score.

Conclusion

ML based diagnosis using two different models showed acceptable predictive accuracy for HFpEF. Limitations of using LVEF as a predictor for the diagnosis of HF by ML and discussion on how the combination of an expert clinician opinion with ML further improves the diagnostic ability will be made. Abstract Figure.

Assessment of nivolumab benefit-risk profile of a 240-mg flat dose relative to a 3-mg/kg dosing regimen in patients with advanced tumors

Background

Nivolumab 3 mg/kg every 2 weeks (Q2W) has shown benefit versus the standard of care in melanoma, non-small cell lung cancer (NSCLC), and renal cell carcinoma (RCC). However, flat dosing is expected to shorten preparation time and improve ease of administration. With knowledge of nivolumab safety, efficacy, and pharmacokinetics across a wide dose range in body weight (BW) dosing, assessment of the benefit–risk profile of a 240-mg flat dose relative to the approved 3-mg/kg dose was approached by quantitative clinical pharmacology.

Patients and methods

A flat dose of 240 mg was selected based on its equivalence to the 3-mg/kg dose at the median BW of ∼80 kg in patients in the nivolumab program. The benefit–risk profile of nivolumab 240 mg was evaluated by comparing exposures at 3 mg/kg Q2W and 240 mg Q2W across BW and tumor types; clinical safety at 3 mg/kg Q2W by BW and exposure quartiles in melanoma, NSCLC, and RCC; and safety and efficacy at 240 mg Q2W relative to 3 mg/kg Q2W in melanoma, NSCLC, and RCC.

Results

The median nivolumab exposure and its distribution at 240 mg Q2W were similar to 3 mg/kg Q2W in the simulated population. Safety analyses did not demonstrate a clinically meaningful relationship between BW or nivolumab exposure quartiles and frequency or severity of adverse events. The predicted safety and efficacy were similar across nivolumab exposure ranges achieved with 3 mg/kg Q2W or 240 mg Q2W flat dose.

Conclusion

Based on population pharmacokinetic modeling, established flat exposure–response relationships for efficacy and safety, and clinical safety, the benefit–risk profile of nivolumab 240 mg Q2W was comparable to 3 mg/kg Q2W. The quantitative clinical pharmacology approach provided evidence for regulatory decision-making on dose modification, obviating the need for an independent clinical study.