MAOMAO

Multi-horizon Anticipatory Outcome Model for Anesthesia and Operations

Research demo

Input JSON

static: age, male (0/1), ASA, emergency (0/1), weight (kg), height (cm).

events: time_min = minutes since record start; token = vocabulary name; optional value = raw measurement.

Dataset

Custom parameters

p = softmax((logits + bias) / temperature)

Presets are source-specific. Fit locally for a new source. Only next-event scores are calibrated; waiting time and horizon outputs remain raw.

#	Event	Raw	Calibrated	Wait (h / min)	1h (raw)	6h (raw)	24h (raw)

Next-event relative probabilities sum to 1; they are not independent concurrent-event risks.