The Hospital Information System (HIS) has been developed based on the action and receipt of patient orders. Recently, various studies and developments have emerged for the education, management, and treatment of patients. Such information systems offer essential information to medical service providers to improve decision-making, and communication between medical service providers and consumers. The direction of medical information has grown toward the improvement of patient satisfaction and high-quality services. Furthermore, the need to provide effective medical information services has increased, which corresponds to the latest information and communications technology, such as mobile and individually customized services [1]. Above all, long waiting times (20.2%) is the main cause of patient dissatisfaction, along with expensive medical bills (24.7%) and unsatisfactory treatments (20.7%) [2,3,4]. From the perspective of patients, waiting time is important when evaluating the quality of medical services [5,6,7,8]. Therefore, almost all medical institutions attempt to find solutions to control the patient service process. This corresponds to the utilization of an appointment policy to control visiting times and scheduling rules for an efficient patient service sequence [9]. Most large hospitals have scheduling rules based on reservation policy and client relationship management (CRM). In addition, hospital staff can effectively manage integrated reservations through scheduling rules [10,11]. Thorough education in medical environments and examination orders are required for reservation counselors. Such education can help to decrease the ratio of patient complaints because patient-centered schedule management depends on the reservationist's skill. The administration of contrast media into the blood vessels should be scheduled when blood tests are taken [12,13]. Therefore, reservation management systems were designed to offer a user-centric user interface (UI) to prevent human error and improve customer satisfaction [14]. First, rules were defined using the sequential pattern of mining techniques through the pattern analysis of exams conducted on each outpatient [15]. Second, a one & nonstop environment through master management of exams and a multi-reservation viewer were designed to manage difficulties and exam constraints [16]. Recommendation algorithms consist of content-based algorithms and rule based algorithms. Content-based algorithms can easily reflect the properties and usages of the recommended subjects. In this study, we used a rule-based algorithm because it could cause problems, such as the complexity of the calculation and inaccuracy of new user and definitions of the terms. There are several types of the rule-based algorithms, such as online convergence detection (OCD), set oriented mining (SETM), direct hashing and pruning (DHP). However, we used an apriori algorithm because it is easy for an apriori algorithm to understand, implement, and generate a proper result to find a set of frequent items in the binary association rules. Because of the small amount of contents and the lack of a need for complicated calculation, reservation and examination in the hospital environment did not require a systemic response in accordance with the pattern. The apriori algorithm was able to make a recommendation in a short time, and this advantage could overcome the limitations of scalability and sparsity. The set of items called market basket analysis could be classified as exams, so it is proper for them to be applied in the examination reservation part of the general hospital environment. However, this study did not focus on the performance of a rule-based algorithm [15,17,18,19,20].

Reservation counselors could easily and rapidly manage their job through HERS. In addition, reservations for patients could be made with reduced waiting times, which had a positive effect on optimized medical services. Most mining technology-based recommendation systems have negative effects when the input values are incorrect. Therefore, their degree of recommendation was not high. Furthermore, methods based on keyword entry require numerous operations that use the weight of relationships between user and content. As a result, their recommendation performance is lower. In this paper, we proposed a recommendation system based on doctor prescriptions to overcome the aforementioned problems. The proposed system removes those exams that were not executed to improve recommendation accuracy. Moreover, we analyzed the patterns of executed exams based on the data of each department in this study. This system supported the ability to extract more accurate patterns by limiting the appropriate range and threshold. In the implementation test, many frequent items that could be used were created because the connections between exams are strengthened. The results section in HERS was limited by the number of items that could be displayed on the (association rules) orders of examination reservations to reference four items. The reason for this limitation was to prevent the concentration phenomenon of particular content with high support value. In this paper, we used an apriori algorithm. By recommending closer dates and faster times, the proposed system could provide effective service to hospital staff and patients. In addition, we proved the accuracy and efficiency of HERS through real operation in a hospital. In the future, we will research medical data mining to further improve the efficiency and accuracy of our service. HERS will be an advanced extension of various applications and services. In addition, the service will be improved to meet users' needs through continuing research.