Early detection of atypical cognitive-motor development is critical for timely intervention, yet traditional assessments rely heavily on subjective, static evaluations. The integration of digital devices offers an opportunity for continuous, objective monitoring through digital biomarkers. In this work, we propose an AI-driven longitudinal framework to model developmental trajectories in children aged 18 months to 8 years. Using a dataset of tablet-based interactions collected over multiple academic years, we analyzed six cognitive-motor tasks (e.g., fine motor control, reaction time). We applied dimensionality reduction (t-SNE) and unsupervised clustering (K-Means++) to identify distinct developmental phenotypes and tracked individual transitions between these profiles over time. Our analysis reveals three distinct profiles: low, medium, and high performance. Crucially, longitudinal tracking highlights a high stability in the low-performance cluster (>90% retention in early years), suggesting that early deficits tend to persist without intervention. Conversely, higher-performance clusters show greater variability, potentially reflecting engagement factors. This study validates the use of unsupervised learning on touchscreen data to uncover heterogeneous developmental paths. The identified profiles serve as scalable, data-driven proxies for cognitive growth, offering a foundation for early screening tools and personalized pediatric interventions.
Figures & tables
Fig. 1: Overview of the longitudinal data collection and processing pipeline used in the study.
Educational Level
Age Range
Course 2
18 Months - 2 Years
Course 3
2-3 Years
Course 4
3-4 Years
Course 5
4-5 Years
Course 6
5-6 Years
Course 7
6-7 Years
TABLE I: Relationship between educational levels (courses) and age ranges.
Fig. 2: t-SNE projection of K-Means++ clustering results for Course 4 based on the six cognitive-motor task scores.
Cluster
Q1
Q2
Q3
Q4
Q5
Q6
%
Course
0
2.39
6.26
0.03
2.34
17.04
29.44
67.31
02
1
52.99
5.07
0.00
3.81
16.51
20.42
32.69
02
0
2.82
5.86
0.50
2.00
16.34
28.04
40.80
03
1
60.60
14.64
4.68
11.88
23.36
36.78
59.20
03
0
2.49
7.37
0.33
2.84
17.96
29.66
25.36
04
1
63.20
15.77
1.21
8.86
26.43
42.90
40.71
04
TABLE II: Average Q scores per cluster and course. The percentage column indicates the distribution of samples in each cluster.
Course Transition
Stable (%)
Improving (%)
Declining (%)
Course 2 → 3
94.1
5.9
0.0
Course 3 → 4
85.7
9.4
4.9
Course 4 → 5
78.6
0.0
21.4
Course 5 → 6
73.2
11.0
15.8
Course 6 → 7
69.4
14.3
16.3
Course 7 → 8
82.0
7.0
11.0
TABLE III: Summary of cluster mobility across course transitions. Percentages indicate the proportion of children whose cluster status remained stable, improved, or declined.
Course Transition
Cluster 0
Cluster 1
Cluster 2
Course 2 → 3
100.0
88.0
–
Course 3 → 4
92.7
80.1
66.6
Course 4 → 5
94.1
77.4
60.0
Course 5 → 6
91.3
68.2
62.4
Course 6 → 7
89.5
70.4
66.1
Course 7 → 8
68.5
95.1
–
TABLE IV: Stability rates (%) of each cluster across course transitions. High percentages reflect consistent developmental profiles year-over-year.