Healthleap scans hospital records for malnutrition risk to help dietitians prioritize

Hospitals miss a lot of malnutrition. Healthleap is trying to fix that by looking at every adult inpatient’s chart each day and dropping a nutrition-risk score into the EHR flowsheet every morning. The point isn’t to replace dietitians. It’s to shove the highest-risk patients to the top of the list before they disappear into routine chart review.

That matters because the gap is big. The source says up to 50% of inpatients are at risk of malnutrition, fewer than 9% are diagnosed, and a typical 1,000-bed hospital can miss about 5,500 malnourished patients a year. Healthleap says it fits upstream of CDI and coding, identifying risk on day one rather than waiting for downstream documentation workflows to catch up.

The company says its malnutrition module was validated against the clinical reference standard in research co-authored with leaders from Cedars-Sinai, Johns Hopkins, and Stanford, and published in Applied Clinical Informatics. It claims 88% higher sensitivity than nurse-administered screening, 4 days earlier identification, and AUROC of 0.95 during stay and 0.92 on day one. It also says that, to its knowledge, this is the only commercially available peer-reviewed AI malnutrition screening tool.

The business case is doing a lot of work here. Healthleap points to Penn Medicine, where it says the system produced a $23.8 million annualized financial impact and 8,632 annualized bed-days saved, and to Cedars-Sinai, where it says impact reached $11 million annually with 39% more diagnoses at the same staffing level. The company adds that every 1,000 additional reimbursable malnutrition cases captured is worth roughly $5 million in incremental reimbursement.

And malnutrition is just the opening act. Healthleap says the same daily, EHR-native screening approach is being extended to delirium and toxic metabolic encephalopathy, aspiration pneumonia, congestive heart failure readmissions, and pressure injuries. Hospitals have spent years paying for more software to make existing workflows louder; this pitch is different because it wants to make the first signal better.