Table 2.

A comparison between our previous studies and this extended study.

Study	Study focus	Dataset used	Recommendation type	Method focus
Chatterjee et al.36	Conceptualized the idea of weekly activity forecasting with statistical models and a rule-base for personalized rule-based recommendation generation in activity eCoaching	PMData	Personalized	ARIMA, SARIMA, Kalman Filter, Rule-database
Chatterjee et al.37	Conceptualized the idea of weekly activity forecasting and a rule-base for personalized recommendation generation with Ontology reasoning and querying in activity eCoaching	PMData	Personalized	LSTM, Ontology
Chatterjee et al.38	Semantic ontology model to annotate the machine learning (ML)-classification outcomes and personal preferences to conceptualize personalized recommendation generation with a hybrid approach in activity eCoaching with a focus on transfer learning approach to improve ML model training and its performance, and an incremental learning approach to handle daily growing data and fit them into the ML models (Support Vector, Naive Bayes, Decision Tree, K-Nearest Neighbour, Random Forest)	Zenodo Fitbit and MOX2-5	Personalized	Standard ML classification models, Ontology
Our work	Design and development of an extended ontology model for semantic representation of personal and personalized activity data, and algorithm development to include time-series forecasting, time-series physical activity level classification, and statistical metrics in the ontology model for hybrid recommendation generation with person-based heuristic configuration and the verification of the algorithm against different datasets with existing and derived metrics	PMData and MOX2-5	Personalized	Deep learning models, Ontology, Probabilistic Interval Prediction, Statistical Metrics