LO
DeploymentCognitive automation2020-01-01
Mount Sinai's dietitians use a deployed machine-learning score at five hospitals to decide which inpatients to assess first for malnutrition
Dietitian / nutritionistoccupation page →
Event date / reported
2020-01-01
Evidence stage
DeploymentAn employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
Tasks this bears on
Assessing what someone eats
Recording and estimating a person's intake from food diaries, recalls and photos.
Being augmented✓ Evidence-backed
Where this applies
New York, United States. The health system's own researchers say it deployed a machine-learning model (MUST-Plus) to detect malnutrition on hospital admission, that the model has been deployed at five of its hospitals since 2020, and that the registered dietitian team uses its prediction to prioritise which inpatients to evaluate on daily rounds, after which the dietitian documents a malnutrition diagnosis against clinical criteria. The paper also finds the model was miscalibrated in some periods and groups. One health system; the model ranks patients for dietitians, it does not assess or diagnose them, and no change in dietitian time is reported.
What this means
In real hospitals, software now decides the order in which dietitians see patients, while the assessment stays with the dietitian.
What it does not yet show
One health system, a model that ranks rather than assesses, and no measure of dietitians' time.
What you can check
Open the npj Digital Medicine paper (PMC11156633) and find "used by the RD team to prioritize which inpatients to evaluate".
Does it change the assessment?
No. The impact index is never moved by a single event. What this record did: the 1 linked task judgement above now rest on evidence instead of inference.
Source
Mount Sinai Health System authors — 'Assessing calibration and bias of a deployed machine learning malnutrition prediction model within a large healthcare system', npj Digital Medicine 7:149 (6 June 2024) · verified 2026-10-09 · Claude (VOLO agent) · interpreted 2026-10-09 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.