Table 2 Data collection framework for marathon training analysis.

From: Machine learning-based personalized training models for optimizing marathon performance through pyramidal and polarized training intensity distributions

Feature category

Specific metrics

Sampling frequency

Data processing

Relevance to training models

Cardiovascular

Heart rate, HRV, HR zones, recovery HR

Continuous (1 Hz)

R-R interval analysis, smoothing algorithms

Primary intensity zone classification

Movement & GPS

Distance, pace, elevation, route mapping

Continuous (1 Hz)

GPS correction, elevation smoothing

Training load quantification

Training load

Session duration, distance covered, elevation gain

Per session

Session summation, weekly totals

Volume quantification

Subjective measures

RPE, fatigue score, sleep quality, readiness

Daily/Post-session

Standardized scales (1–10)

Recovery status indicator

Performance testing

Time trials (5 km, 10 km, half-marathon), VO₂max

Bi-weekly/Monthly

Laboratory protocols

Model validation metrics

Environmental

Temperature, humidity, wind conditions

Per session

Weather integration

External load modifier

Biomechanical

Cadence, stride frequency (from GPS watch)

Continuous (1 Hz)

Digital filtering

Running efficiency indicator