AI-BASED E-HEMAP: AN LLM-SUPPORTED SMART COST MANAGEMENT AND PROACTIVE DECISION SUPPORT SYSTEM FOR HEALTHCARE-ASSOCIATED INFECTIONS

Abstract

The invention relates to the technical fields of Health Management, Medical Informatics, and Artificial Intelligence; it concerns e-HEMAP (Electronic Hospital Infection Cost Analysis Program), an expert decision support system that enables the simultaneous analysis of structured data within Hospital Information Management Systems (HIMS) and unstructured clinical data such as physician / nurse summaries, consultation and report texts using Natural Language Processing (NLP) and Big Language Models (LLM); the proactive prediction of Healthcare-Associated Infection (HAI) risk before clinical symptoms and diagnostic criteria emerge; and the reporting of direct and indirect additional costs associated with these infections at macro and micro levels by separating them from the patient's main diagnostic costs at the automation level. The invention also relates to the system's working method. The system in question comprises an Input Layer (100) where clinical data is collected, a Communication Layer (110) which ensures secure data transmission, a Data Lake Architecture (120) where data is stored, and a Processing Layer (130) that performs semantic scanning with an LLM architecture. The system also includes an NLP Analysis Unit (210) that distinguishes negative expressions in free text, a Filtering Unit (220) that clarifies parameters by filtering data, and ICD-10, SNOMED-CT, TC.The system includes the Matching / Evaluation Unit (230), which matches the Ministry of Health National SHIE Surveillance Diagnostic Guide with SUT, CDC / NHSN diagnostic criteria; the Decision Mechanism (240), which generates proactive alerts before clinical signs and diagnostic criteria appear using dynamic scoring and rule-based logic; the Cost Analysis Module (250), which automatically separates the costs of additional drugs, laboratory / radiology tests, consultations, medical consumables and materials, other interventional procedures and additional hospitalization days from the main diagnostic costs; and the e-HEMAP Cost Analysis Component (350). The system also includes the Billing Unit (440), which prepares consolidated invoice printouts, and the Management and Approval Mechanism (490), where data is submitted for EKK approval and external system integrations are provided.

Description

1 TARIFF AI-BASED E-HEMAP: HEALTHCARE-RELATED SMART COST-EFFECTIVE TREATMENT OF INFECTIONS BASED ON LARGE TONGUE MODEL. MANAGEMENT AND PROACTIVE DECISION SUPPORT SYSTEM Technical Area 5 The invention relates to the technical fields of Health Management, Medical Informatics, and Artificial Intelligence; furthermore... specifically structured data within Hospital Information Management Systems (HIMS). including physician / nurse summaries, free-text clinical notes, and consultation reports. Natural Language Processing (NLP) and Big Language Models (LLM) of unstructured clinical data Simultaneous analysis using these methods, even before clinical signs and diagnostic criteria emerge. firstly, proactively assessing the risk of Healthcare-Associated Infections (HSAI / hospital infections) anticipating and determining the additional direct and indirect costs arising from this infection The patient's main diagnostic costs are automatically separated at the macro and micro levels. e-HEMAP (Electronic Hospital) is an expert decision support system that enables reporting. It is related to the Infection Cost Analysis Program and methodology. 15 Previous Technique Detection and monitoring of SHIE in secondary and tertiary level hospitals providing healthcare services. For this purpose, Hospital Information Systems (HIS) and Electronic Health Records have been used for many years. In these systems... Infection Control Committees (ICCs) based on consultation requests from clinics. Surveillance activities are carried out by [organization name]. In the current practice, EKKs (Electronymic Coordination Centers) assess clinical conditions. After it was formed, the International CDC (Centers for Disease Control and Prevention) / NHSN According to the National Healthcare Safety Network criteria and the guidelines of the Turkish Ministry of Health. By verifying the case, we only submit patient information, diagnosis and type of SHIE to the system / Ministry. This process is entirely reactive and works retrospectively (looking back to the past); The current job description and system infrastructure of the EKKs (Electrical Counseling Committees) enable them to guide the treatment process of SHIE (School of Emergency Medicine), 25 This does not directly involve calculating or billing for additional costs. With the development of artificial intelligence, machine learning, and natural language processing technologies in recent years With the advent of electronic systems, decision support systems have begun to become widespread in the healthcare field. Analyzing clinical data found in health records to estimate the risk of SHIE, 2 various software programs to support patient classification and clinical decision-making processes. Solutions have been developed. These systems mostly include laboratory values, drug records, and diagnoses. Codes and similar structured data are analyzed using specific algorithms. The probability of SHIE (Surprise Identification) is calculated. In some applications, natural language processing techniques are used. While it is possible to analyze physician notes, these analyses are mostly limited to certain 5 This is done based on the identification of keywords. In current technical solutions, SHIE surveillance is largely based on structured clinical data. It is based on free-text clinical notes entered into the system by physicians and nurses. Semantic analysis is not widely practiced. This is included in clinical notes. Contextual expressions, negative suffixes, temporal relationships, and clinical meaning differences are sufficient. 10 Since it could not be accurately assessed, important findings regarding the early stages of SHIE were overlooked. It can go unnoticed. Therefore, SHIE is often diagnosed after the clinical picture has developed. This is possible, and the systems operate using a reactive approach. Rule-based algorithms used in current systems to determine SHIE risk, High percentage of 15 because they could not fully interpret the context of the medical terms in the text. It can produce false positive and false negative results. Especially "fever not monitored", "discharge". Negative statements or temporal descriptions such as "no" or "infection was not suspected" are not accurate. Because it could not be evaluated, it caused SHIE control teams to encounter unnecessary warnings. It is possible. In addition, existing hospital information systems and decision support systems have an additional 20% capacity due to the SHIE (School Emergency Management Authority). automatic separation of costs from treatment costs related to the main diagnosis is unable to provide. The medications administered to the patient, laboratory tests, imaging procedures, Which of the consultations, increased hospital stay, and other interventional procedures are due to SHIE? The emergence of these problems is mostly determined through manual assessments. This situation leads to costs. significant time loss in analyses, subjective evaluations and reporting 25 This leads to mistakes. The current technology involves a central data architecture that gathers data from different healthcare organizations. Large language models (LLMs), natural language models, enable common analysis by combining them under one umbrella. Processing (NLP), machine learning algorithms, and SHIE cost analysis all on the same platform. Decision support systems that integrate these elements are not widely available. Furthermore, SHIE 30 Estimating the risk before a clinical diagnosis is made, and comparing this risk with international SHIE criteria. 3 Automatic evaluation by comparison and cost-related expenses of SHIE within the same system. Simultaneous calculation within it is not sufficient in current solutions. It cannot be accomplished. Document CN118280564A describes how artificial intelligence can be used to identify patients at high risk of sepsis. 5 based on clinical data that enables the identification and early warning of these patients. It describes a system that performs analysis and prediction. However, the document in question states that... structured and unstructured clinical data combined in a large language model (LLM) contextual analysis using [method], identification of SHIE risk and additional SHIE-related risks A solution for automatically separating costs from main diagnostic costs. It is not explained. 10 Document CN117936075A describes how sepsis risk can be interpreted based on monitoring data. to analyze through models and generate early warning outputs It discusses the method. However, this document does not mention the surveillance criteria specific to SHIE. by evaluating the additional costs arising from SHIE, these costs are automatically deducted from the main diagnostic costs. An integrated decision support system for its decomposition is not described. 15 Document US11398301B1 describes the risk of sepsis by analyzing patient data. determination and clinical decision support outcomes based on the results obtained It refers to a system for its creation. However, in this system, SHIE with contextual analysis of unstructured clinical data based on Big Language Model (LLM) a system that automatically separates the source costs together 20 It is not explained. Document CN121506416A describes the analysis of patient data using multimodal data fusion. and support decision-making processes in an intensive care setting It refers to the system. However, this document mentions SHIE's international surveillance criteria. Accordingly, the assessment and additional costs arising from SHIE are automatically deducted from the main diagnostic costs. No mechanism for separating these components is explained. In document US2022084662A1, medical intervention for a patient is determined by analyzing patient data. to determine the need in advance and to create early warning accordingly. It refers to a system. However, the document in question discusses the identification of SHIE risk. 4 Automatic separation of additional costs arising from SHIE from main diagnostic costs. No integrated solution providing this is described. In document US12254966B1, data obtained from doctor-patient interactions and patient information are included. Analysis of clinical data using large language model (LLM) based artificial intelligence techniques. and as a result of this analysis, diagnostic, recommendation and decision support outputs are provided to healthcare professionals. 5 A system for presenting [LLM] is mentioned. However, the document in question refers to LLM. linking the base analyses with the surveillance criteria specific to SHIE and additional SHIE-derived data. An integrated decision support system that performs cost analysis is not described. Existing patent documents relating to artificial intelligence-based clinical decision support systems A significant portion of this involves determining the risk of sepsis or the clinical course of intensive care patients. It is aimed at prediction. However, the concepts of sepsis and SHIE have technical and clinical aspects. They differ from each other in terms of aspects. Sepsis is a type of infection that includes various infections, including community-acquired infections. While other conditions can develop as a systemic clinical syndrome, SHIE is present at the time of hospital admission. non-existent and developing during the delivery of health services, national and international surveillance It refers to infections evaluated according to the criteria. Therefore, each sepsis case is 15. Not every case of SHIE (Ship-Infected Infection) is of a similar nature, and not every SHIE case develops into sepsis. This Therefore, sepsis-focused decision support systems utilize the SHIE's unique surveillance criteria. by evaluating the additional costs arising from SHIE, these costs are automatically deducted from the main diagnostic costs. It does not solve the problem of separation. Current technical solutions analyze both structured and unstructured clinical data together. 20 by proactively predicting the risk of SHIE, using large language models and clinical notes. capable of performing semantic analysis, automated decision-making in accordance with international SHIE criteria. can offer support and automatically deduct the additional costs incurred due to SHIE from the main treatment costs. We fully understand the need for an integrated decision support system that can report by separating it into categories. This situation is not met by natural language processing and semantics using large language models. the analysis layer, contextual filtering mechanism, SHIE criteria matching module, proactive risk scoring algorithm and SHIE-derived cost analysis module on the same platform. what brings them together and, most importantly, the additional costs associated with this infection, is the main diagnosis. a new integrated system that automatically separates costs and reports them at the macro / micro level This reveals the need for a decision support system (e-HEMAP). 30 Purposes and Brief Description of the Invention The purpose of this invention is related to the technical fields of health management, medical informatics, and artificial intelligence; More specifically, structured data obtained from Hospital Information Management Systems (HIMS). data including physician / nurse summaries, free-text clinical notes, and consultation reports. Natural Language Processing (NLP) and Big Language Models 5 of unstructured clinical data Simultaneous analysis using (LLM) when clinical signs and diagnostic criteria have not yet been established Proactively predicting the risk of SHIE before it occurs and the resulting complications from this infection. The resulting direct and indirect additional costs are automatically deducted from the patient's main diagnostic costs. An expert decision support system that enables the analysis and reporting of data at both macro and micro levels. with e-HEMAP (Electronic Hospital Infection Cost Analysis Program) and its method 10 It is related. Another aim of the invention is to enable natural language processing (NLP) and large language model (LLM) based analysis. Using these methods, the negative aspects of medical terms found in free-text clinical notes can be examined. by ensuring accurate assessment of appendices, temporal relationships and clinical context The goal is to reduce false positive and false negative results in SHIE (Self-Emission Testing). 15 Another aim of the invention is to predict the risk of developing SHIE before a clinical diagnosis is made, thereby improving healthcare. The goal is to provide proactive decision support to professionals and enable early intervention. Another purpose of the invention is to reduce the additional costs arising from SHIE (drugs, medical supplies / materials, laboratory / radiology tests, additional hospital stays and other interventional procedures); Health In accordance with the Ministry's National SHIE Surveillance and Diagnostic Guidelines standards, national, regional and 20 Automated, objective data at the provincial level, based on hospital, clinic, physician, patient, and diagnosis / infection type. and to report by analyzing and parsing in a dynamic way. Detailed Description of the Invention e-HEMAP, a large language model for SHIE, was developed to achieve the objectives of this invention. (LLM) supported intelligent cost management and proactive decision support system, as shown in the attached figures 25 It has been shown. Figure 1: Overall system of the e-HEMAP decision support system supported by a large language model (LLM). It is the architecture and workflow diagram. 6 The parts shown in the figures are individually numbered, and their corresponding numbers are listed below. It has been given. 100 – Input layer 110 – Communication layer 5 120 – Data lake architecture 130 – Processing layer 210 – NLP analysis unit 220 – Filtering unit 230 – Matching / evaluation unit 10 240 – Decision-making mechanism 250 – Cost analysis module 350 – e-HEMAP cost analysis component 440 – Billing Unit 490 – Management and approval mechanism 15 Healthcare management is related to the technical fields of medical informatics and artificial intelligence, more specifically. with structured data obtained from hospital information management systems (HIMS) the computer-based analysis of unstructured clinical data, large language Using the model of contextual clinical meaning (LLM) and natural language processing (NLP) techniques 20 to make inferences, to determine the risk of SHIE and to determine the costs associated with this infection e- Analysis using HEMAP (electronic Hospital Infection Cost Analysis Program) computer-based decision support method for, - Structured data obtained from hospitals (laboratory results, etc.) unstructured data (such as doctor and nurse notes, radiology / pathology reports, etc.) (free texts) are transferred to the system, - Using natural language processing (NLP) and big language modeling (LLM) on the transferred data to be analyzed, - Contextual evaluation of clinical statements within textual data and interpretation, 30 - The analyzed data were used in medical classification systems (ICD-10, SNOMED-CT, SUT) (Health Practice Regulation) Matching with CDC and NHSN, etc. 7 - Determining and estimating the risk of SHIE based on patient data, - Warning and decision support outputs from the system depending on the SHIE risk. creation, - Identifying and analyzing cost items related to SHIE (medication, length of stay, tests, etc.). done, 5 - A combined evaluation of the clinical and financial data obtained, - System outputs displayed as reports, alerts, or analyses for healthcare professionals. presented through It includes the steps. The system that is the subject of the invention, 10 - Enables the collection of clinical data (structured and unstructured data) input layer (100), - A communication system that ensures the secure transfer of collected data to a central system. layer (110), - A data lake architecture that enables the consolidation and storage of collected data. 15 (120), - Semantic analysis of clinical texts through the integration of the Large Language Model (LLM) and a processing layer that performs the interpretation (130), - An NLP that enables the analysis of clinical texts using natural language processing methods. Analysis unit (210), 20 - A system that removes erroneous, incomplete, or irrelevant information from the data. Filtering unit (220), - Enabling the matching of clinical data with international criteria. Matching / Evaluation Unit (230), - A decision that enables the determination of the risk of SHIE based on patient data 25 mechanism (240), - A cost analysis tool that enables the analysis of cost elements related to SHIE. module (250), - Enables the consolidation and invoicing of costs related to SHIE (State Economic Enterprises). billing unit (440), 30 8 - An e-HEMAP cost analysis that enables the separation of additional costs arising from SHIE. component (350), - A management and approval system that ensures system integration and process verification. mechanism (490) It includes 5. The system enables the collection of clinical data (structured and unstructured data). an entry layer (100), ensuring the secure transfer of collected data to the central system a communication layer (110) that enables data to be consolidated and stored data lake architecture (120), semantics of clinical texts with large language model (LLM) integration a processing layer (130) that performs the analysis and interpretation of clinical texts natural 10 an NLP analysis unit (210) that enables the analysis of data using language processing methods, a filtering unit that enables the removal of erroneous, incomplete or irrelevant information (220), a matching / evaluation that enables the matching of clinical data with international criteria Unit (230) is a decision that enables the determination of the risk of SHIE based on patient data. mechanism (240) is a cost 15 that enables the analysis of cost elements related to SHIE. Analysis module (250), consolidation and invoicing of SHIE related costs. a billing unit (440) that provides the separation of additional costs arising from SHIE. System integration and processes with an e-HEMAP cost analysis component (350) It includes a management and approval mechanism (490) that ensures its verification. Before the system started operating, the hospital information management system (HIMS), laboratory 20 Integration with information management systems and other health information systems is ensured. Thanks to integration, laboratory results, vital signs, medication records, and imaging can be accessed. results and similar structured clinical data by doctors and nurses Free text generated clinical notes, radiology / pathology reports input layer (100) It is received into the system via. The collected data is securely stored by the communication layer (110) 25 It is transferred to a central data lake architecture (120) using communication protocols. Data lake architecture (120), clinical data obtained from different hospitals and health organizations It enables the consolidation of data into a central data structure. Thus, different Data on patient groups with socioeconomic and demographic characteristics are available on a single platform. It is being analyzed, and the large language model and other artificial intelligence algorithms are testing different data 30 9 This allows the system to work on clusters. Thanks to its central data structure, the system It can be made scalable in a way that allows it to create a common analytical infrastructure across different healthcare organizations. is being brought in. Data stored in the central data lake architecture (120) is transferred to the processing layer (130). Processing layer (130), structured clinical data and unstructured free text 5 preparing the records for the analysis process and the Big Language Model (LLM) and NLP analysis unit (210) enables them to work together. NLP analysis unit (210), doctor / nurse observation notes, radiology / pathology reports By applying natural language processing techniques, we can identify disease findings and symptoms within the text. laboratory terms, treatment information, drug names, and other medical concepts 10 This determines the large language model that runs in the processing layer (130). It is semantically evaluated by [the relevant authority]. The large language model does not only perform keyword matching, but also in clinical texts. It also analyzes the context of the statements used. In this context, "no current", "fire not detected", Negative suffixes in expressions such as "post-operative third day" reflect time relationships and clinical 15 The context is considered together. Thus, it is not based solely on the identification of specific words. False positive and false negative assessments resulting from traditional methods are significant. is being reduced to a certain extent. Data semantically evaluated by LLM are filtered by unit (220) is transferred. The filtering unit (220) detects missing, erroneous, duplicate or clinical evaluation 20 It improves the accuracy of the analysis by filtering out irrelevant data that could have a negative impact. Similarly... time filters and negative expressions in time, temporal relationships and clinical context Only usable in SHIE assessment by applying relevant contextual controls. Clinical information is included in the analysis process. Clinical data that have completed the filtering process are sent to the matching / evaluation unit (230) 25 Clinical concepts determined in this unit are based on ICD-10, SNOMED-CT, and Turkish Health Regulations. The Ministry's National SHIE Surveillance Diagnostic Guide, SUT, CDC and NHSN, and other national and It is compared with internationally accepted SHIE criteria. The defined clinical concepts matching / evaluation unit (230) with the aforementioned SHIE criteria They are compared, and as a result of this comparison, a criterion fit coefficient of 30 is obtained for each patient. This criterion is being created. This created fit coefficient is used in determining the risk of SHIE (School-Based Impact Assessment). It is used. Laboratory results, vital signs, medication administration, and the coefficient of agreement with the criteria. Clinical observation notes are evaluated together by the decision-making mechanism (240). Decision mechanism (240) calculates the probability of developing a healthcare-associated infection and 5 This risk of SHIE in the patient before significant clinical signs and diagnostic criteria emerge It can predict the risk in the previous latent (precursor) phase. Pre-determining the risk level. This enables healthcare professionals to receive early warnings and necessary clinical advice. This ensures that interventions can be planned before SHIE develops. Decision-making mechanism Automatic alarm and decision support for patients identified as high risk by (240) 10 Outputs are being generated, and the identified SHIE risk is being visualized for healthcare professionals to make decisions. It is presented in its wooden structure. As a result of the risk assessment created by the decision mechanism (240), the system, the relevant It creates a digital follow-up record for the patient on the e-HEMAP platform. (SHIE criteria) If it is not provided at a sufficient level, the registration is closed as negative, and the SHIE risk is 15. If verification or identification of high risk occurs, the record is kept active and a cost analysis is performed. The process is initiated. Thus, detailed information is only collected for confirmed or high-risk cases. A cost assessment is being carried out. Decision mechanism (240), criteria compliance obtained from matching / evaluation unit (230). coefficient, laboratory / radiology results, vital signs, drug administrations and clinical 20 By evaluating the observation notes together, the system determines the patient's risk of SHIE (School-Related Infection). Risk If the level is above a certain threshold value, the cost analysis module (250) e-HEMAP cost analysis is triggered and used to calculate costs arising from SHIE. component (350) is being put into operation. Cost analysis module (250) 25 depending on the results obtained from the decision mechanism (240) It is activated. The cost analysis module (250) analyzes the costs incurred due to SHIE. e-HEMAP works together with the cost analysis component (350) for the purpose of determining. HEMAP cost analysis component (350) includes drug expenses incurred due to SHIE, antibiotics their uses, laboratory and radiological examinations, consultations, medical supplies and materials, increased hospital stay, and other interventional procedure cost items are added to the main diagnosis at 30 11 It automatically separates the related treatment costs. Thus, additional costs associated with SHIE are deducted. Costs can be calculated objectively and financial analysis can be performed. Examples of other interventional procedure cost items originating from SHIE analyzed by the system. This is shown in Table 1. Table 1. Other Interventional Procedure Cost Items 5 No Cost Items No Cost Items 1 IV injection 16 Drug administration via nebulizer 2. Catheter dressing and care 17. Intravenous access 3 Monitoring (daily) 18 Total parenteral nutrition (TPN) monitoring 4 Intravenous drug infusion 19 Wound dressing Emergency hemodialysis / intensive care and bedside hemodialysis 20 Bladder catheterization 6 Cardiopulmonary resuscitation 21 Central venous catheterization, jugular or subclavian 7 Cardioversion 22 Intraarterial cannulation + pressure measurement 8 Colostomy care 23 Endotracheal tube or tracheotomy cannula change 9. Follow-up of enteral hyperalimentation; 24. Patient education for COPD patients. Subcutaneous injection 25 Inhaler device training 11-day inpatient pain monitoring, 26 oxygen inhalation therapy sessions. 12 Steam therapy 27 Transthoracic echocardiography 13 Wound debridement and dressing (major) 28 Electrocardiogram 14 Nasogastric tube insertion 29 Decubitus ulcer dressing Endotracheal intubation (outside the operating room) 30 Blood or product transfusion e-HEMAP cost with clinical evaluation results obtained from the decision mechanism (240) Cost analysis results generated by the analysis component (350) are correlated with the same patient 12 They are evaluated and reported together under the record. The prepared clinical and financial evaluation results billing unit (440) via relevant hospital information management The system is being transferred to financial management systems or other enterprise software. At the same time... Data sharing with different software infrastructures can be carried out through API services, and The system is designed to work in an integrated manner across different healthcare facilities. 5 Records, reports, cost analyses, and patient evaluations are generated by the system. The results are verified by the management and approval mechanism (490). Management and approval mechanism (490) supports the case review processes of the SHIE control committee, The system ensures the verification of records and carries out the necessary approval processes. It increases its reliability. 10 In this context, the data are distributed as follows: input layer (100), communication layer (110), data lake architecture (120), processing layer (130), NLP analysis unit (210), filtering unit (220), Matching / Evaluation Unit (230), Decision Mechanism (240), Cost Analysis Module (250), e-HEMAP cost analysis component (350), billing unit (440) and management and approval It is processed through the mechanism (490). 15 This process is repeated for each patient, and structured and unstructured clinical data are collected. joint assessment of healthcare-associated infections before clinical diagnosis predicting, automatically allocating, and managing the costs associated with this infection, health providing early warning and decision support outputs to professionals and controlling SHIE It ensures that processes are managed in an integrated manner, both clinically and financially. 20 In a sample application of the invention, the patient presents to the clinic providing the treatment and with the main diagnosis... After the patient is admitted, clinical data is continuously monitored by the system. It is being monitored. Healthcare-related data is being tracked by an AI-based decision support system. If a suspected infection (SHIE) is identified, a record is created for the patient on the e-HEMAP platform. It is being created and the presence of SHIE is being assessed. Determination that SHIE is not present 25 If this is the case, the record is closed; however, if SHIE is confirmed, the patient's treatment process is monitored. These infection-related costs are analyzed through the e-HEMAP cost analysis component. The costs are calculated. The calculated costs are transferred to the relevant billing processes, and the resulting costs are... The results obtained are evaluated by the SHIE control committee and subjected to necessary verification and approval. The processes are carried out. Thus, the invention, clinical decision support, SHIE monitoring and cost 30 It manages analysis processes in an integrated manner.

Claims

13 REQUESTS 1. It relates to the technical fields of health management, medical informatics, and artificial intelligence, more specifically. With structured data within Hospital Information Management Systems (HIMS) including physician / nurse summaries, free-text clinical notes, and consultation reports. Natural Language Processing (NLP) and Big Language Models (LLM) of unstructured clinical data 5 Simultaneous analysis using these methods, before clinical signs and diagnostic criteria emerge. first, the risk of Healthcare Service-Associated Infections (HSAIs) must be proactively predicted and this additional direct and indirect costs arising from infection the patient's main diagnosis Reporting at macro and micro levels by automatically separating costs. e-HEMAP (Electronic Hospital Infection Control 10), an expert decision support system, provides It is related to the Cost Analysis Program and its methodology; - Structured clinical data and unstructured clinical data obtained from hospitals. transferring data to the system, - Large language model (LLM) of transferred structured and unstructured clinical data and analysis using natural language processing (NLP) techniques, 15 - The contextual analysis of medical terms found in the clinical texts examined. evaluation of negative statements, temporal relationships and clinical Interpretation of context, - The clinical concepts obtained are based on ICD-10, SNOMED-CT, and the Turkish Ministry of Health National Data Collection. The SHIE Surveillance Diagnostic Guide, SUT, CDC and / or NHSN SHIE criteria are in line with 20 comparison and, as a result of this comparison, a criterion of suitability for each individual. creation of the coefficient, - The established criteria fit coefficient together with structured clinical data. Identifying the risk of SHIE by evaluating it, - Creation of warning and decision support outputs for patients identified as at risk of SHIE, 25 - Cost items related to SHIE (medication, length of hospital stay, tests, etc.) should be considered in relation to the main diagnosis and treatment. Analyzing it by separating it from its costs, - Clinical evaluation results and cost analysis results are correlated and presented together. reporting, - The generated results are presented to healthcare professionals as reports, alerts, or analysis screens. 30 presented through 14 It is characterized by including its steps.

2. The working method described in Claim 1 is natural language processing (NLP) of unstructured clinical data. analyzed using techniques and semantically by the large language model (LLM) It is characterized by its evaluation.

3. The study method in Claim 1 is based on the large language model (LLM) and is included in clinical texts. 5 the field of negative expressions, temporal relationships and clinical context together It is characterized by its evaluation.

4. The study method in Claim 1 is as follows, and the clinical concepts analyzed are from ICD-10, SNOMED-CT, Turkish Ministry of Health National SHIE Surveillance Diagnostic Guide, SUT, CDC and / or NHSN. Comparison with SHIE criteria and, as a result of this comparison, criterion compliance for each patient is 10. It is characterized by the creation of a coefficient.

5. The working method in Claim 1 is as follows, and the established criterion fit coefficient is laboratory / radiology. results, vital signs, medication administration, and clinical observation notes should be evaluated together. By determining the risk of SHIE (School-Related Infection) and based on the determined infection risk, automated... It is characterized by the generation of decision support outputs based on alarms. 15 6. The working method is as described in Claim 1, and the identified SHIE risk is communicated to healthcare personnel. It is characterized by being presented in a visualized decision tree structure.

7. The working method in Claim 1, and drug expenses related to SHIE, antibiotic use, laboratory and radiological examinations, consultations, increased hospital stay, interventional procedures, medical e-HEMAP cost analysis component 20 of consumables and materials and other cost items It is characterized by being separated from treatment costs related to the main diagnosis using this method.

8. The study method described in Claim 1 is based on clinical evaluation results and cost analysis. reporting the results together, linked under the same patient record, and related information. It is characterized by being transferred to their systems.

9. Structured data obtained from hospital information management systems (HIMS) (laboratory 25 results, vital signs, etc.) and clinical data (doctor and nurse observation notes, (such as radiology / pathology reports, free texts) are analyzed together, and from this data Artificial intelligence for contextual clinical interpretation and early detection of SHIEs (School-Related Immunodeficiency Tests). It is a decision support system based on - Enables the collection of clinical data (structured and unstructured data) input layer (100), - A communication system that ensures the secure transfer of collected data to the central system. 5 layer (110), - A data lake architecture that enables the consolidation and storage of collected data. (120), - Semantic analysis of clinical texts through the integration of the Large Language Model (LLM) and a processing layer that performs the interpretation (130), 10 - An NLP that enables the analysis of clinical texts using natural language processing methods. analysis unit (210), - A system that removes erroneous, incomplete, or irrelevant information from the data. Filtering unit (220), - A 15 that enables the matching of clinical data with international criteria. Matching / Evaluation Unit (230), - A decision that enables the determination of the risk of SHIE based on patient data. mechanism (240), - A cost analysis tool that enables the analysis of cost elements related to SHIE. module (250), 20 - Enables the consolidation and invoicing of costs related to SHIE (State Economic Enterprises). billing unit (440), - An e-HEMAP cost analysis that enables the separation of additional costs arising from SHIE. component (350), - A management and approval system that ensures system integration and process verification. 25 mechanism (490) It is characterized by its inclusion.

10. The AI-based decision support system is as in claim 9, and the input layer is (100), hospital information management system, laboratory information management system and other health information by collecting structured and unstructured clinical data from its systems 30 It is characterized by... 16 11. It is an AI-based decision support system as in Claim 9, and the communication layer (110), Clinical data received from the input layer (100) are collected into the data lake using secure communication protocols. It is characterized by its transfer to its architecture (120).

12. The decision support system is an AI-based system, as in Claim 9, and the processing layer (130), By running the big language model (LLM) and the NLP analysis unit (210) together, 5 clinical texts It is characterized by its ability to perform semantic analysis.

13. The decision support system is an AI-based system, as in Claim 9, and the filtering unit (220), filtering out erroneous, incomplete, duplicate, or irrelevant information from clinical data and contextual controls regarding negative statements, temporal relationships, and clinical context. It is characterized by its application. 10 14. An AI-based decision support system, as in Claim 9, for matching / evaluation. The unit (230) analyzed clinical data according to ICD-10, SNOMED-CT, TC Ministry of Health. Comparison of the National SHIE Surveillance Diagnostic Guide with SUT, CDC and / or NHSN criteria. It is characterized by establishing a criteria fit coefficient for each patient.

15. The decision support system is an artificial intelligence based system as in claim 9, and the decision mechanism is (240), 15 along with the criteria fit coefficient, laboratory results, vital signs, medication administration, and By evaluating clinical observation notes together, the risk of SHIE is determined, and this identified SHIE... It is characterized by presenting the risk to healthcare professionals in a visualized decision tree structure. is being done.

16. As in Claim 9, it is an AI-based decision support system, and the cost analysis module is 20. (250), working together with the e-HEMAP cost analysis component (350), drug damage due to SHIE expenses, antibiotic use, laboratory and radiological tests, consultations, increased length of hospital stay, invasive procedures, medical supplies and equipment, and other cost items It is characterized by separating the treatment costs from the main diagnosis.

17. As in Claim 9, it is an AI-based decision support system, and the billing unit is 25 (440), financial outputs related to the cost analysis results created, hospital information management The system is characterized by its transfer to financial management systems or other enterprise software. is being done. 17 18. As in Claim 9, it is an AI-based decision support system for management and approval. the mechanism (490), records, reports, cost analyses and generated by the system by verifying the patient evaluation results and carrying out the necessary approval procedures. It is characterized by... 10 20

Images