Extended Data Fig. 3: Disease progression tracking using a different number of nights.

a, b, 6-month and 12-month change in MDS-UPDRS as assessed by a clinician and predicted by the AI model, both using a single night and multiple nights of data. On each box, the central line indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The whiskers extend to 1.5 times the interquartile range. Similar to the clinician assessment, when using only a single night of data, the AI model cannot detect statistically significant changes over a year (p = 0.751 for 6 months, p = 0.235 for 12 months, one-tailed one-sample Wilcoxon signed-rank test). This indicates that the reason why the AI model can achieve statistical significance for progression analysis while MDS-UPDRS cannot stems from being able to combine measurements from multiple nights, which substantially reduces measurement noise and increases sensitivity.