AI-Based Diagnostic Test Prediction- An Effective Way to Handle the New Normal Era in Healthcare

Disease Diagnosis (DD) healthcare recommender systems can significantly contribute to disease control by accurately predicting disease risk and aiding in disease management. These systems serve as must-be supplementary tools for physicians and patients, assisting them in effectively controlling and managing diseases. However, the development of accurate real-time recommendations for medical data poses contests due to the complex nature of the data, which is often unbalanced, large, multi-dimensional, noisy, and may contain lost values. The effectiveness of diagnostic tests (DT) is known to vary based on the context in which they are utilized, such as their position in the clinical pathway. In practice, tests are typically employed alongside other pertinent information, including patient age, sex, symptoms, clinical signs, and the results of other tests, to facilitate accurate diagnoses. Recognizing the interplay of these factors is pivotal for achieving optimal diagnostic outcomes. DTs are an essential aspect of ensuring patient safety and effective healthcare. However, in the post-pandemic condition, patients have become more cautious about visiting hospitals frequently, as they aim to minimize exposure to contagious diseases like SARS-CoV-2 infection. Consequently, there is a growing preference among patients to limit hospital visits. To address this concern, the integration of deep learning (DL), machine learning (ML), and other artificial intelligence (AI) techniques has proven to be highly effective in analyzing large medical datasets and recognizing patterns. AI-based diagnostic systems hold promise in accurately predicting diseases based on patients' symptoms, thereby supporting existing clinical practices. In this work, an AI-based recommender scheme was industrialized to suggest primary DTs using patient symptoms. Data were collected through a questionnaire from patients in the Kolhapur district of Maharashtra, India, and verified by medical professionals. To ensure data quality, an Extract, Transform, and Load (ETL) algorithm was employed for preprocessing. The collected data revealed significant associations between symptoms and probable diseases, enabling the system to provide relevant recommendations for primary DTs. For both patients and clinicians, predictive standards and their relative to result thresholds play a vital role in decision-making. These values depend on the personal preferences and priorities placed on outcomes, including positive and negative results. By considering these factors, the developed AI-based recommender system aims to assist in informed decision-making, offering accurate disease predictions while taking individual values into account.