Healthleap’s $38 Million Raise Targets Missed Diagnoses Inside the Hospital EHR
Healthleap has raised $38 million in combined seed and Series A financing as it pushes a clinical AI platform deeper into hospital workflows. The San Francisco company is built to screen hospitalized patients for conditions that may otherwise go undiagnosed, beginning with malnutrition and extending next to congestive heart failure, aspiration pneumonia, delirium, pressure ulcers, and selected hospital-acquired infections.
The financing came from Sequoia Capital, First Round Capital, and Hummingbird Ventures. For a hospital market squeezed by staffing shortages and rising costs, the appeal is straightforward: identify risk earlier, triage it faster, and do it without adding another layer of alert fatigue.
Malnutrition Is the First Use Case, and It Fits the Weakest Points in Usual Screening
Healthleap’s current focus is malnutrition, a condition that can quietly worsen recovery, extend hospitalization, and complicate discharge planning. The workflow is designed around a problem hospitals know well: the patients most likely to be missed are often the same patients least able to answer a traditional screening tool. Confusion, intubation, and critical illness can shut down the usual bedside interview, leaving the risk invisible until weight loss, deconditioning, or poor intake becomes harder and more expensive to reverse.
The platform runs inside the electronic health record and screens every adult inpatient daily. Instead of relying on a single admission-time questionnaire, the model pulls together weight, body mass index, laboratory results, diagnoses, medications, intake and output, nutrition orders, and vital signs. Those inputs are consolidated into a priority score from 0 to 10 for each patient, with a second column showing whether the risk changed since the last assessment.
For dietitians, the clinical utility is immediate. The score appears in the existing EHR worklist each morning, where staff can sort patients and start with the highest-risk cases. In practical terms, the system is trying to move the problem from a late-stage rescue situation to a daily operational task.
A Daily Worklist Replaces Interruptive Alerts
The platform does not fire pop-ups at physicians or nurses. It does not interrupt care with another alert demanding a click-through. Instead, the risk score sits in the same worklist clinicians already use.
That design choice matters because hospitals have learned that an algorithm can be technically accurate and still fail at the bedside if it creates friction. A passive score is easier to absorb into dietitian workflow than a noisy alert that competes with medication orders, discharge tasks, and urgent clinical calls. By keeping the system inside the chart review process, Healthleap is betting that adoption depends less on novelty than on workflow compatibility.
“Every condition runs on the same nightly chart review and the same EHR connection,” co-founder and clinician Jemima Meyer said in an email. “Healthleap does not send interruptive alerts. Physicians and nurses see no pop-ups. The score appears as a column in a worklist that the care team already uses.”
Clinical Validation, Reimbursement, and the Economics of Earlier Detection
Healthleap says its platform is already used by more than 50 hospitals, including Penn Medicine, Houston Methodist, Cedars-Sinai, Emory, and UMass Memorial. The company’s website links to a peer-reviewed study supporting a claim of 88% higher sensitivity than nursing screening for malnutrition detection.
The economics around malnutrition are difficult to ignore. Each severe malnutrition diagnosis and treatment supports better patient outcomes and, according to Healthleap, can allow hospitals to claim around $11,000 in reimbursement. Meyer also said that each flagged patient had a 1.6-day shorter average hospital stay. Those figures give hospital leaders a financial rationale for investing in a detection system that may catch nutrition problems before they become irreversible complications.
Malnutrition is only the starting point. The planned expansion into congestive heart failure, aspiration pneumonia, delirium, pressure ulcers, and hospital-acquired infections points to a broader clinical pattern: many inpatient complications are preceded by subtle data changes that are easy for humans to miss when chart volume is high. A rising oxygen requirement, a changing medication profile, falling mobility, or a worsening nutrition trajectory may not trigger a crisis on their own, but together they can signal a patient who is sliding toward avoidable harm.
The Next Test Is Whether the Model Works Across Other Hospital Problems
Extending the same platform to new conditions will require more than simply swapping one diagnosis for another. Each clinical domain has its own signatures, its own confounders, and its own threshold for acceptable false positives. Congestive heart failure looks different from delirium; pressure injury risk is not measured the same way as aspiration risk; hospital-acquired infection detection requires a different balance between sensitivity and specificity than nutrition screening does.
Validation will therefore be the key gatekeeper. As Healthleap expands, each condition must undergo further clinical validation while still fitting into provider workflow without creating new burdens. A model that performs well in malnutrition but struggles in another inpatient syndrome could quickly lose credibility with frontline teams who need trustworthy prioritization, not more digital noise.
For hospitals, the underlying promise is attractive because it is operational as much as clinical. Earlier recognition of overlooked illness can preserve function, shorten stays, and reduce the downstream costs that follow missed diagnoses. For patients, the benefit is simpler: treatment starts before the condition becomes harder to reverse.
The unresolved question is not whether hospitals need better screening. It is how many different inpatient syndromes can be identified reliably enough by the same data-driven approach before validation, specificity, and workflow burden set the limits.