AI for Healthcare: Diagnosis, Clinical-Trial Matching, and Patient Recruitment
Abstract
Medical diagnosis is the most critical component in the treatment of a patient. But diagnosis often is a complicated process since a myriad of diseases share the same symptoms. If a patient is diagnosed with a disease in its end-stage, potential new treatments (clinical trials) are sometimes the last option available. However, matching a patient to the correct clinical-trial requires advanced medical knowledge on behalf of the patient. In this study, we try to address the following problems and close the technical gaps, (i) Diagnosis: Advances in neural network approaches and the availability of massive labeled datasets have sparked renewed interests in automated diagnosis. We explore novel techniques to identify pathology in chest radiographs by using a labeled radiograph dataset, which is also substantially large for the domain of medical diagnosis. (ii) Clinical-Trial Matching: Given the difficulty of perusing the jargon in standard clinical trial texts, we try to complement the process by using machine learning and information retrieval methods to fetch similar health records showing the entities responsible for the match. We implement an efficient visual tool (TextMed) to aid our algorithm and make it easier for users to utilize the power of machine learning. Our tool helps in searching through a database of criteria and records and fetches the information about the query.
Subject
Information RetrievalNatural Language Processing
Computer Vision
Healthcare
Medicine
Bioinformatics
Citation
Das, Karimi Abhishek (2020). AI for Healthcare: Diagnosis, Clinical-Trial Matching, and Patient Recruitment. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /192242.