Fig. 4: Challenge results for Task 3: trajectory segmentation and characterization. | Nature Communications

Fig. 4: Challenge results for Task 3: trajectory segmentation and characterization.

From: Objective comparison of methods to decode anomalous diffusion

Fig. 4: Challenge results for Task 3: trajectory segmentation and characterization.The alternative text for this image may have been generated using AI.

a Root mean square error (RMSE) as a function of the changepoint location along the trajectory for all teams in 1D (teams E, J, and B). Dashed lines represents the RMSE associated to a random prediction of the changepoint position. b Corresponding means absolute error (MAE) of the prediction of α and, (c) F1-score for the classification of the diffusion model for the first (solid symbols/continuous line) and second segment (empty symbols/dashed line) as a function of the changepoint location along the trajectory. d Plot of the recall vs the FPR for all participating teams. e Plot of JSC vs RMSETP for all participating teams. For the calculation of the metrics in d-e, only trajectories presenting a changepoint at a distance larger than 20 points from the start/end points were considered as undergoing a switch. RMSETP was estimated only for true positive position pairs. Colors indicate teams, following the same color code as in a. f, g RMSE as a function of the anomalous diffusion exponent (f) and of the diffusion model (g) of the first and second segment for the best-in-class team in 1D (team E). All results for T3 in 1D, 2D, and 3D are provided in Supplementary Fig. 5.

Back to article page