Medical big data are rapidly being generated and accumulated throughout the healthcare industry. Using such medical data to extract meaningful information is expected to improve healthcare services significantly. In this paper, we present an integrated dataset, consisting of electronic health records (EHRs) and mobile personal health records (mPHRs), which enables high-accuracy disease diagnostics. An EHR represents the overall health status of a patient, including the patient’s past medical records, while mPHRs are data recorded by an individual’s mobile devices and provide real-time health information that varies over time. Accordingly, each EHR and mPHR plays a complementary role in diagnosing a patient’s disease, enabling accurate health diagnostic services. To generate an integrated dataset comprising EHR and mPHR, two tasks are performed for each individual dataset: data preprocessing and data matching. The former includes a formatting step to set the appropriate format for the data and a cleansing step to replace missing values and outliers with median values. The latter task requires overwriting or combining individual attributes within the EHRs and mPHRs into a unified form. For a comparative analysis of the integrated datasets, we generate a prediction model for heart disease using the decision tree method. The results show that the prediction model for the integrated dataset exhibits higher accuracy than individual datasets in predicting a patient’s disease.