/Hip_Implant-from-vGRF-

Primary LanguageMATLAB

Classification of Gait Cycle Abnormalities due to Hip Surgery and Implanted Hip using Machine Learning Techniques

This repo contains the inference code for the paper titled "Classification of Gait Cycle Abnormalities due to Hip Surgery and Implanted Hip using Machine Learning Techniques" shich is submitted to IEEE Access.

This repo has two parts: Result Evaluation and Inference on New Data.

The calculation of classification metrics have been done using statsOfMeasure.m from Mathwork File Exchange

Result Evaluation

In this part, we will load the test data from Data folder and get the prediction by using the models saved in Model folder. The Matlab script result_evaluation.m performs this task.

This snippet of code performs the prediction:

%% Load Test Data and Model; Predict on Test Data
warning('off')
cm = zeros(3,3);
for fold = 1:5
    model_name = sprintf("Model/Fold%d_KNN_model.mat",fold);
    data_name = sprintf("Data/Fold%d_Data.mat",fold);
    load(model_name)
    load(data_name)

    [pred, score, ~] = predict(model.clf, testX);
    cm = cm + confusionmat(testY,pred);
end

Note that we have provided test data and model for each fold.

Inference on New Data

In this part, we will show how to use the new vGRF signals (like those in Sample Data folder) and get predictions using our model.

The script inference_on_new_data.m will perform the inference. This script is a wrapper around inference.m which does the feature extraction and prediction.

The following snippet performs the inference:

%% Test on New Data
% sig: GRF signals in cells of shape (N,1) where each row has signals of
% shape (101,1)
% FS: Sampling Frequency
tic
pred = inference(sig, FS);
toc