Develop a machine learning approach for anomaly detection using similarity measures in complex chronic disease patient data to improve quality issues related to manual data entry.
- We make recommendations for possible values based on predictive models to the clinicians who can either accept or decline the recommendations. In summary, retrospective data of chronic disease patients will be used for anomaly detection problems using supervised, unsupervised and semi-supervised approaches. The historical data will be used to train the supervised machine learning model and will serve as the gold standard.
- The common machine learning algorithms for anomaly detection may include Support Vector Machines (SVM), Multi-Layer Perceptron (MLP) and k-Nearest Neighbour (kNN) with local outlier factor. Their performance will be evaluated using a predefined testing framework to identify the best-performing methods. The best method will be used as an effective decision support tool for the respective chronic condition.