Table 2 MLP and deep RNN classification rates.

The classification validation results are presented in three scenarios (rows) for both models, with nine conditions (columns) for the MLP model and seven conditions (columns) for the deep RNN model. Scenarios: (1) binary classification of healthy-patient individuals, (2) binary classification of affected-unaffected legs, (3) three-class classification of severity grades. MLP conditions: The model is trained and tested only with the extracted features under: (1) entire, (2) male, (3) female, (4) other age, (5) middle-aged, (6) elderly, (7) normal BMI, and (8) overweight individuals data. Also, (9) the model is trained and tested using all extracted features and general risk factors (sex, BMI, and age) under the entire data. Deep RNN conditions: The model is trained without general risk factors and tested with (1) unseen trials and (2) perfect unseen trials. The model is trained with general risk factors and tested with (3) unseen trials and (4) perfect unseen trials. Finally, the model is trained with general risk factors , and its sensitivity to general risk factors is tested using perfect unseen trials in the absence of (5) age, (6) BMI, and (7) sex data.

Search