Artificial intelligence and women’s health

Abstract

The use of artificial intelligence (AI) in medicine is currently an issue of great interest, especially with regard to the diagnostic or predictive analysis of medical data. As information technology in health care continues to develop, greater amounts of clinical medical data have been generated and will continue to expand every year. Diagnostic data, clinical trial data and data on medical staff behavioral health together form the largest data group. AI technology is now capable of deriving algorithms which can be used for diagnosis and treatment of disease and medical research. AI already has learning and self-correcting abilities and may further improve its accuracy based on feedback. An AI system can assist physicians by providing up-to-date medical information from journals, textbooks and clinical practices to inform proper patient care. Additionally, AI may help to reduce diagnostic and therapeutic errors which are inevitable in human clinical practice. AI systems extract information from large patient populations, allowing them to provide real-time advice on health risks and also predict health outcomes. AI devices are categorized into two major groups. The first group includes machine learning techniques that analyze structured data in an attempt to cluster patients’ traits, and consequently predict the probability of disease outcomes. The second group includes natural language processing methods that extract information from unstructured data, such as clinical notes and medical journals, to supplement and enrich structured medical data. The natural language processing procedures turn texts into machine-readable structured data, which can then be analyzed by machine learning techniques. AI is being employed in the medical field in at least four distinct ways: (1) in the assessment of risk of disease onset and in estimating treatment success prior to initiation; (2) in an attempt to manage or alleviate complications; (3) to assist with patient care during the active treatment or procedure phase; and (4) in research aimed at elucidating the pathology or mechanism of and/or the ideal treatment for a disease. The forecasting of health outcomes in different body systems (i.e. risk assesment for any disease) has been extensively investigated. Cardiovascular, breast, bone, cervix and endometrium have been the areas of interest in AI research in women’s health. Artificial neural networks (ANNs) and classification and regression trees for the prediction of endometrial cancer in postmenopausal women have been investigated. Similarly, AI has estimated the impact of human papillomavirus types in influencing the risk of cervical dysplasia recurrence. Extensive use of AI has been used in breast imaging where images from mammographic, sonographic and magnetic resonance imaging (MRI) were collected for study. The feasibility of automatically identifying normal digital mammography examinations with AI to reduce the reading workload of breast cancer screening was examined and it was found that incorporating an AI-based decision support system into ultasound image analysis improved diagnostic performance. Predictive models for osteoporosis have been constructed based on popular machine learning algorithms such as support vector machines, random forests, ANNs, and logistic regression based on simple surveys. These machine learning models were later compared to four conventional clinical decision tools: the Osteoporosis Self-assessment Tool, the Osteoporosis Risk Assessment Instrument, the Simple Calculated Osteoporosis Risk Estimation, and the Osteoporosis Index of Risk. Likewise, the application of an ANN in optimizing the Osteoporosis Self-Assessment Tool for Asians score has been reported. Moreover, it was identified that there are several methods in the use of AI to help the screening of groups at risk for osteoporosis or fractures. However, such systems were limited to a specific ethnic group, gender or age. Furthermore, a performance comparison of machine learning algorithms in predicting fragility fractures from MRI data was conducted where, among many classifiers, the random undersampling-boosted trees, logistic regression, and the linear discriminant are best for predicting osteoporotic fracture. Moreover, morphological, topological and mechanical bone features using AI methods were investigated. In a clinical trial, the performances of the Adaptive Neuro Fuzzy Inference System, support vector machines and genetic algorithms in classifying two populations of arthritic and osteoporotic bone samples were compared. Finally, AI was