Understanding Disease Co-Occurrence On Chest Radiographs Using Exploratory Data Analysis

Chest X-ray imaging is a fundamental diagnostic tool in the evaluation of different thoracic diseases. The concept of co-occurrence patterns of diseases in chest X-ray reports is pivotal to clinical decision-making, predictive analysis, and treatment planning. This study utilized the CheXpert dataset, a large-scale dataset of chest radiographs with annotations, to explore disease co-occurrence patterns. Data preprocessing techniques were used to handle uncertain labels and missing values in the dataset. The cleaned dataset was used to compute the co-occurrence matrix, which quantifies the relationships between different diseases. In addition, statistical methods and visualization techniques, such as heatmap was used to elucidate these relationships. The findings of this study revealed remarkable correlations and associations between specific diseases, providing insights into disease comorbidities and potential diagnostic considerations. Also the clinical implication of the result for diagnosis and patient management was reported. Therefore, this study contributed to the existing body of knowledge and research in chest radiograph analysis for effective medical imaging analysis and clinical practices. Keyword: chest radiographs, EDA, missing values, heatmap, correlation, comorbidity Olawuyi, Nkechinyere Joy & Asunbiaro, Charles Sileola (2025): Understanding Disease Co-Occurrence On Chest Radiographs Using Exploratory Data Analysis. Journal of Advances in Mathematical & Computational Science. Vol. 13, No. 2. Pp 67-79. Available online at www.isteams.net/mathematics-computationaljournal. dx.doi.org/10.22624/AIMS/MATHS/V13N2P6