Leveraging process mining for modeling progression trajectories in amyotrophic lateral sclerosis

Abstract

Background: Amyotrophic Lateral Sclerosis (ALS) is a neurodegenerative disease whose spreading and progression mechanisms are still unclear. The ability to predict ALS prognosis would improve the patients' quality of life and support clinicians in planning treatments. In this paper, we investigate ALS evolution trajectories using Process Mining (PM) techniques enriched to both easily mine processes and automatically reveal how the pathways differentiate according to patients' characteristics.

Methods: We consider data collected in two distinct data sources, namely the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) dataset and a real-world clinical register (ALS-BS) including data of patients followed up in two tertiary clinical centers of Brescia (Italy). With a focus on the functional abilities progressively impaired as the disease progresses, we use two Process Discovery methods, namely the Directly-Follows Graph and the CareFlow Miner, to mine the population disease trajectories on the PRO-ACT dataset. We characterize the impairment trajectories in terms of patterns, timing, and probabilities, and investigate the effect of some patients' characteristics at onset on the followed paths. Finally, we perform a comparative study of the impairment trajectories mined in PRO-ACT versus ALS-BS.

Results: We delineate the progression pathways on PRO-ACT, identifying the predominant disabilities at different stages of the disease: for instance, 85% of patients enter the trials without disabilities, and 48% of them experience the impairment of Walking/Self-care abilities first. We then test how a spinal onset increases the risk of experiencing the loss of Walking/Self-care ability as first impairment (52% vs. 27% of patients develop it as the first impairment in the spinal vs. the bulbar cohorts, respectively), as well as how an older age at onset corresponds to a more rapid progression to death. When compared, the PRO-ACT and the ALS-BS patient populations present some similarities in terms of natural progression of the disease, as well as some differences in terms of observed trajectories plausibly due to the trial scheduling and recruitment criteria.

Conclusions: We exploited PM to provide an overview of the evolution scenarios of an ALS trial population and to preliminary compare it to the progression observed in a clinical cohort. Future work will focus on further improving the understanding of the disease progression mechanisms, by including additional real-world subjects as well as by extending the set of events considered in the impairment trajectories.

Keywords: Amyotrophic lateral sclerosis; Patient stratification; Process discovery; Process mining; Prognosis; Progression trajectories.

Fig. 1

DFG graph representing the paths followed by the study population, delineating the increase in disability experienced by the subjects. Only the arcs with a transition probability $>0.03$ are shown

Fig. 2

a Zoom on the DFG deltaGraph obtained stratifying the population by onset site (spinal vs. bulbar). The highlighted edges represent an increased transition probability for the spinal (red) or bulbar (green) cohort, thresholded for displaying only differences between the probabilities greater than 0.2. b KM curves of the time passing from M_0000 (no impaired domains) to M_1000 (Walking/Self-care domain impaired) for the two cohorts. The log-rank test shows statistically significant differences between the cohorts. $+$ indicates censored subjects

Fig. 3

CFM graph built starting from M_0000. Each node reports the total number of patients passing through it (round brackets) and the min-median-max time, in days, needed to reach it from the root. Colors are graded on the median times, with intervals: <100, 101-200, 201-400, and >401 days. The edges report the percentage of patients passing through the child node with respect to the previous node (above) and the entire population (below). The graph has been thresholded for displaying only the pathways transitioned by at least 10 subjects

Fig. 4

CFM graphs built starting from M_0000 and stratified for a quantized age at onset or b onset site. Each node reports the number of patients passing trough it for each cohort (young/aged onset and spinal/bulbar onset, respectively). Moreover, the ratio of the two cardinalities with respect to the initial populations is reported in the round brackets, followed by the p-value for the Fisher’s exact/${\chi }^2$ test. The node box is colored in yellow if the p-value is lower than a given threshold (here 0.05). Both the graphs have been thresholded for displaying only the pathways transitioned by at least 10 subjects

Fig. 5

On the top, a portion of the DFG showing the differences between PRO-ACT and ALS–BS in terms of probability to move from a state to the next one is shown. The gray edges represents the transitions whose probabilities differ for less than 0.2 in the two datasets. When this gap is larger, the edge is green if ALS–BS has a probability higher than PRO-ACT, red otherwise. On the bottom, the kernel density estimation of the probability to move (a) from M_1000 to M_0101; (b) from M_1000 to M_1110, and (c) from M_1001 to M_1111 are reported, with PRO-ACT in blue and ALS–BS in red, and with the vertical dotted line indicating the median of the distributions