With the rapid popularization of the Internet of Healthcare Systems (IHS), the scale and concentration of medical data have increased significantly, and the value has also been greatly improved. However, the high dependence on technology has also made the risk of technology-dependent data increasingly severe. At present, the industry mainly relies on expert assessment for risk assessment, and there is still a lack of flexible and effective technical methods for medical data risk assessment in the environment of the IHS. Therefore, a medical data risk assessment method based on improved evidence theory and Analytic Hierarchy Process (AHP) is proposed. This method includes the whole process of medical data risk assessment index system design, risk index weight calculation and risk assessment. Aiming at the problems of fuzziness, uncertainty and conflict in expert evaluation, a fusion calculation method of subjective and objective weight of medical data risk indicators based on AHP-CRITIC method and an improved evidence theory based on Pignistic probability distance and conflict coefficient \(K\) are proposed. Finally, through empirical analysis, the medical data risk value of a hospital information system is calculated as (0.0355, 0.1269, 0.1871, 0.4935, 0.1570), indicating a low risk level, which is consistent with the actual situation, which proves the effectiveness of the proposed method. The scientific rigor and accuracy of the improved method are substantiated through comparative experimental analysis.