Voice electronic medical record system based on artificial intelligence and block chain technology

Abstract

The invention discloses a voice electronic medical record system based on artificial intelligence and a block chain technology, relates to the technical field of electronic medical record systems, and is characterized in that the system comprises a medical record management module, a medical record authority management module, a text encryption and decryption module, a block chain client module, a wallet management module and a login management module. Text content is extracted from the doctor-patient dialogue, and error correction is performed on the text content to determine the accuracy of the text content. Necessary information of the electronic medical record is extracted from the corrected text by using a natural language processing technology, so that a preliminary abstract is generated. The physician may further add refined medical data by making manual reviews and modifications using their clinical acuity, or activate an artificial intelligence diagnostic support module to supplement evidence-based suggestions. The processed data provided by the doctor is stored on the block chain, and encryption is applied to ensure strong data security and privacy.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic medical record systems, and more specifically, to a voice electronic medical record system based on artificial intelligence and blockchain technology. Background Art

[0002] The digital transformation of healthcare is underway, driven by the acceleration of technological advances that are revolutionizing how healthcare is delivered and how data is leveraged in health systems around the world. Key information technologies such as electronic health records, telemedicine, and mobile health applications have connected patients and providers in unprecedented ways, enabling more effective care coordination, lower costs, and improved health outcomes. As emerging economies invest in healthcare infrastructure and developed countries seek to control spending, digital health solutions have become a key strategic tool for enhancing system performance. Cloud computing, 5G networks, artificial intelligence, and the Internet of Things offer tremendous potential for extracting insights from the vast amounts of health data generated every day around the world. Data analytics and artificial intelligence are already uncovering new medical knowledge that will power the next generation of personalized, precision medicine. For these reasons, implementing integrated, digitized models of care has become a pressing priority for healthcare stakeholders. The global digital health market is expected to grow at nearly 19% annually, from $211 billion in 2022 to $800 billion in 2030. Countries that invest in digital health systems now will realize dividends for generations to come in the form of healthier populations, more sustainable healthcare spending, and medical innovations that drive economic growth. The digital transformation of healthcare is inevitable, necessary, and imperative—an opportunity that must be vigorously embraced by healthcare providers, policymakers, and patients around the world.

[0003] The integration of voice recognition into electronic medical records (EMRs) is rapidly transforming clinical documentation while accelerating the digital transformation of healthcare. The ability for physicians to dictate patient access directly into the EMR via voice, rather than manually entering notes, can increase productivity, improve documentation quality, and allow physicians to focus their valuable time on patients rather than administrative burdens. Studies have shown that voice EMRs improve note completion speed, physician satisfaction, and positively impact patient interactions compared to traditional wording or writing methods. In addition to reducing clinician workload, voice EMRs can also facilitate enhanced clinical workflows, improve care coordination across teams, and more fully capture data to power next-generation analytics and decision support tools. As artificial intelligence continues to refine the accuracy of voice recognition, voice EMRs have tremendous potential to strengthen clinical-patient relationships, information flow, and overall quality of care. With its scalability across large healthcare networks, voice EMRs have become a strategic asset for healthcare systems pursuing digital transformation. Voice technology has officially become an integral component of digital health infrastructure, helping providers increase productivity, reduce burnout for time-pressed clinicians, and deliver superior, technology-enhanced patient care.

[0004] The implementation of blockchain technology offers transformative benefits for managing and securing electronic medical records (EMRs) across diverse healthcare organizations. By distributing EMR storage on an encrypted, immutable ledger across multiple nodes, blockchain overcomes data silos and lack of interoperability that have previously hindered effective health information exchange. This enables true portability and continuity of health records, allowing patients to seamlessly share their complete medical history with any provider, not limited to proprietary EMR systems. Blockchain’s cryptographic foundation enhances security and prevents unauthorized tampering or falsification of records. Every transaction is verified and time-stamped through a decentralized consensus mechanism, creating an indelible audit trail for improved data accountability and integrity. Granular access controls using public / private key cryptography ensure that patients can independently decide who has access to their personal health information. These inherent features of blockchain architecture significantly reduce the privacy and security risks of centralized EMR databases. Health systems no longer need to attempt to transfer, reconcile, and manage external records. Patients benefit from continuous care and can instantly get more information on test results, treatment processes, medications, and any provider they see. By intrinsically building patient privacy, ownership, and data integrity into its distributed, cryptographically secure framework, blockchain technology promises to fully unlock the potential of EHRs. Blockchain-powered EHRs can transform fragmented islands of health data into connected, trusted, and transparent records that remain under patient control while flowing seamlessly across a global clinical information network.

[0005] Therefore, this study developed a voice electronic medical record system based on artificial intelligence and blockchain technology. The designed process extracts text content from doctor-patient conversations and corrects it to determine its accuracy. Natural language processing technology is then used to extract the necessary information of the electronic medical record from the corrected text to generate a preliminary summary. Doctors can further augment the refined medical data by manually reviewing and revising it using their clinical acumen, or activate the artificial intelligence diagnosis support module to supplement the evidence-based recommendations. The processed data provided by the doctor is stored on the blockchain, and encryption is applied to ensure strong data security and privacy. Summary of the invention

[0006] The purpose of this invention is to provide a voice electronic medical record system based on artificial intelligence and blockchain technology to enhance clinical documentation and care coordination.

[0007] The above technical objectives of the present invention are achieved through the following technical solutions: a voice electronic medical record system based on artificial intelligence and blockchain technology, the system comprising a medical record management module, a medical record authority management module, a text encryption and decryption module, a blockchain client module, a wallet management module and a login management module.

[0008] The present invention is further configured as follows: the process of generating electronic medical records in the medical record management module is as follows:

[0009] (1) Program relevant clinical questions, collect responses through a conversational user interface, extract and summarize keywords and sentences from these responses to populate the electronic medical record, and collect basic patient information, including age, gender, medical history, and other details required for care delivery;

[0010] (2) Obtain specific consultation information through doctor questions, patient answers, orderly diagnosis and prescription treatment, and then use the Chinese voice network to transcribe the complex voice conversation information between doctors and patients into text data, and use natural language processing (NLP) technology to process these text data and organize them into a structured format;

[0011] (3) Manual verification and modification by physicians: This feature allows physicians to review and modify automatically generated patient information to ensure the accuracy of the data and to add annotations to imaging reports or physiological indicators from patient examinations.

[0012] The present invention is further configured such that patients control their own electronic medical records through the medical record authority management module, retaining the ability to view and authorize, but lacking the ability to modify or add new records.

[0013] The present invention is further configured that: the patient writes registration or login information through the login management module;

[0014] The doctor accesses the system by entering an identification number and password through the login management module, and subsequently accesses a list of registered patients and creates a medical record indexed for the patient;

[0015] Administrators manage the list of doctors in their respective departments through the login management module.

[0016] The present invention is further configured such that: the regional chain client realizes the interaction between patients, doctors, managers and blockchain.

[0017] The present invention is further configured such that: the text encryption and decryption module encrypts and decrypts the medical record content to protect the privacy of the patient.

[0018] The present invention is further configured such that: the wallet management module is used to pay the patient's medical expenses.

[0019] In summary, the present invention has the following beneficial effects: The effective aggregation of clinical data by the system is conducive to the optimization of disease diagnosis, hospital management and nursing workflow. By integrating emerging technologies such as artificial intelligence and big data analysis, combining innovation management with digital strategies to overcome related challenges, the blueprint for digital health is steadily realized.

[0020] The paper-based workflow of hospitals in the prior art is inefficient, takes up a lot of physical space, and hinders the ability to search and share information. The system provided by the present invention enhances the existing electronic medical record model through artificial intelligence algorithms, simplifies voice-to-text transcription, and realizes paperless records. Paperless records are an important milestone in the digitalization of hospitals. It is conducive to the comprehensive intelligent management of hospitals in the future and the establishment of sound digital medical records in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is the Chinese voice network model architecture of the present invention;

[0022] Figure 2 It is the electronic medical record dynamic management platform of the present invention. DETAILED DESCRIPTION

[0023] The following is combined with Figure 1-2 The present invention is described in further detail.

[0024] Example 1: Building a Chinese Voice Network Model

[0025] Previous research in the field of speech-to-text conversion has mainly focused on English speech recognition, while Chinese speech recognition is more difficult due to the tonal nature of the language, the large number of homophones, dialect and accent variations, the complexity of characters, the lack of spaces between words, and the possible lack of high-quality annotated data for certain dialects or accents. Significant progress has been made in English automatic speech recognition (ASR) systems, but Chinese speech-to-text conversion remains an ongoing challenge due to the inherent complexity of tonal variations and homophones in the language. To address this issue, we propose a new model called Chinese Voice Network that combines the strengths of Graph Neural Network (GNN) and Wave Neural Network models, paving the way for more accurate and reliable Chinese speech-to-text conversion systems.

[0026] like Figure 1 As shown in the figure, the Chinese speech network architecture extracts mel frequency reciprocal coefficients (MFCCs) from the raw audio waveform. The MFCC sequence is preprocessed by an initial convolutional layer to obtain a robust input embedding. The adaptive adjacency matrix modeled by the graph convolutional network captures the relationship between acoustic features. A series of dilated convolutional layers inspired by WaveNet effectively propagate information and learn higher-level feature representations. Each convolutional block contains parallel gate and filter convolutions with residual and skip connections to enhance the representation capacity. Additional convolutional and activation layers transform the features into Chinese character predictions.

[0027] We curate a dataset of real patient-doctor consultation conversations as high-fidelity training material that closely reflects real-world applications. The construction of this dataset requires a rigorous half-year process to accumulate these valuable conversation samples. To ensure diversity and representativeness, we recruited 5 doctors to participate in data collection, and each doctor devoted 8 hours a day to consultation. This arrangement not only enriches the breadth of the dataset, but also covers a variety of medical conditions, doctor styles, and patient needs. To ensure the quality and accuracy of the dataset, we carefully screened and pruned external clips. The finalized dataset has a total of 1,200 hours of audio recordings. This comprehensive dataset construction provides rich, realistic speech data, which effectively enhances the training and performance enhancement of our system.

[0028] To optimize performance, we used data augmentation techniques during training. We artificially expanded the dataset by applying small perturbations to the original audio, such as adding background noise or simulating different room acoustics. This exposed the model to greater acoustic variability and improved generalization. We also utilized spectrogram masking, where parts of the input spectrogram are randomly masked, forcing the network to rely on contextual information. In addition, we applied a dropout model to the convolutional layers. These augmentation and regularization methods prevent overfitting and enhance robustness. By combining representation learning, architecture optimization, and rigorous training methods, China Voice Network achieves state-of-the-art speech recognition accuracy.

[0029] The combination of doctor-patient conversation transcripts requires careful processing and correction to improve the accuracy of subsequent speech analysis and processing. Given the medical nature of these conversations, they are likely to contain a large amount of specialized terminology. To accommodate this, we developed a comprehensive medical lexicon in addition to a traditional natural language lexicon. The lexicon includes general vocabulary, specialized medical terms, and drug nomenclature. Its compilation is informed by specialized terminology extracted from a wide range of medical literature. We further improved the accuracy of text correction by incorporating this medical lexicon into the training of a BERT text correction model. This fine-tuning enables the model to skillfully correct medical text. When the model encounters a match, it replaces the medical term in the source text with an exact equivalent based on the definition or suggested alternative vocabulary in our medical lexicon. This correction procedure improves the accuracy of transcribed text from doctor-patient conversation transcripts and ensures the correct use of medical terminology. By implementing these methods and procedures, we provide a sophisticated, improved solution for speech processing and text correction in the medical field.

[0030] Example 2: Generation of electronic medical records

[0031] As healthcare systems navigate the accelerated pace of the digital age, the transformative shift from paper records to digital platforms has become central, with electronic medical records (EMRs) being the cornerstone of this digital healthcare revolution. Improved efficiency, patient care, and cost-effectiveness are among the many benefits of innovative digital-era strategies that streamline EMRs to facilitate remote access, organized data storage, and collaborative decision making. The transition to structured EMRs, where patient information is categorized into specific fields, has further optimized healthcare delivery and data management. This structured organization enables rapid retrieval and interpretation of data, supports decision-making efficiency, and is more conducive to advanced data analytics and interoperability than unstructured EMRs. The standardized format of structured EMRs allows for seamless data exchange with other digital systems, promotes integrated care across diverse healthcare settings, and provides valuable insights through large-scale data analytics, predictive modeling, and machine learning applications. Integrating voice systems into structured EMRs brings further benefits, revolutionizing the efficiency and accessibility of healthcare. Voice recognition software quickly transcribes spoken words into text, reduces manual data entry, and captures complex healthcare medical terminology and patient narratives with improved accuracy. The technology enhances patient engagement, improves accessibility, and increases overall efficiency and cost-effectiveness in healthcare environments. In order to align with the healthcare industry’s digital era innovation strategy, enhance patient care experience, and facilitate data analysis for scientists, we developed the following solutions to generate structured electronic medical records. The specific process is summarized below.

[0032] Generating a structured electronic medical record (EMR) is a multi-step process that combines cutting-edge technology with meticulous data collection and verification practices. The initial step is to transcribe complex voice information into text data using the Chinese Voice Network. This text data is then processed using Natural Language Processing (NLP) technology, which can effectively understand human language and organize it into a structured format.

[0033] The data collection process includes two main components. First, basic patient information is collected, including age, gender, medical history, and other details required for care provision. Since this information is structured, it can be effectively captured through basic human-computer interaction. Responses are collected using a conversational user interface through programming of relevant clinical questions. Keywords and sentences are then extracted and summarized from these responses to populate the electronic medical record. Second, information about a specific consultation is obtained, including physician inquiries, patient responses, ordered diagnoses, and prescribed treatments. Given the uniqueness of each case, a more flexible approach is required. Natural language processing technology is used to analyze and structure patient-provider conversations, extract valid information, and elicit treatment recommendations. In this way, heterogeneous consultation data is systematically converted into structured inputs for the electronic medical record. On the one hand, structured demographic data is seamlessly collected through scripted human-computer interaction, and on the other hand, unstructured consultation information is intelligently extracted using NLP methods. This two-pronged workflow enables detailed patient records to be comprehensive and efficient.

[0034] The physician's inquiries include descriptions of symptoms, medical history, allergies, medication use, etc., while the patient's responses provide relevant clinical information and health context. Through these conversations, the system can automatically infer potential diagnoses and treatment decisions, which are subsequently encoded into structured electronic medical record entries. In addition, the prescribed treatment plan constitutes a complete consultation component. The physician can provide targeted treatment recommendations, drug prescriptions, or surgical options, which require accurate documentation for reference and execution. Through natural language processing, the system identifies and extracts the physician's recommendations on the treatment plan, converting it into a structured representation to enhance understanding and enable the medical team to simplify implementation. In summary, by leveraging automatic speech recognition and natural language processing, unstructured recordings of patient-provider conversations can be converted into structured medical concepts and treatment plans, accurately populating key fields in the electronic medical record. This helps provide effective and comprehensive documentation to inform patient care.

[0035] When generating structured EMRs, a key step to ensure the accuracy and reliability of the data is introduced: manual verification and modification by doctors. This feature allows doctors to review and modify automatically generated patient information to obtain data accuracy. It is also possible to add annotations to imaging reports or physiological indicators from patient examinations.

[0036] These structured EMRs not only facilitate collaboration and communication among healthcare professionals, but are also a valuable resource for data analysis and research. By combining advanced technologies such as speech-to-text conversion and NLP with a thorough validation process, high-quality, reliable and structured EMRs can be generated, which are indispensable in patient-centered care delivery in the digital age.

[0037] Visual Representation of Structured EMR Interface: The electronic medical record (EMR) will be presented through a web interface to provide users with easy access. The page consists of two main parts - a left menu and a right content panel. The persistent left menu bar features a variety of options, including symptoms, physical examination results, diagnosis, laboratory tests, imaging studies, treatments, medications, procedures, and surgeries. Each entry has a checkbox to toggle the display. Users can customize the visible content according to their needs, which improves browsing efficiency. The right panel displays the EMR in a clear format. First, the header displays core record details such as institution, payment method, health card, and record number to quickly identify the patient. Second, personal information including demographic data provides comprehensive background information. Past medical history is recorded in a table format to summarize past conditions. In addition, physical examination contains comprehensive data to facilitate complete patient evaluation. Our system extracts key consultation details and automatically populates the EMR to ensure accurate and complete documentation. Finally, editable "Submit" and "Edit" buttons allow modification and finalization of automatically generated content.

[0038] The intuitive EMR web interface facilitates seamless access and management for healthcare providers and patients. It accurately presents patient data to inspire clinical decisions, while customization improves user workflow efficiency. Supplemental documentation demonstrates the system’s effectiveness in balancing automated documentation and user-friendly visualization to provide an optimal EMR experience. This blend of automation and usability is designed to streamline recordkeeping for more informed, effective patient care.

[0039] Example 3: Blockchain-based file storage and access technology further improves the electronic medical record system

[0040] According to the hospital's operational logistics, including internal department management, patient registration and doctor consultation procedures, combined with the existing electronic medical record model, the electronic medical record system is further developed based on blockchain technology. The system can effectively maintain the privacy of patients and facilitate remote consultation and online treatment by doctors.

[0041] The solution introduces a structured authoritative management system to provide skilled and secure technical assistance to doctors who need to access or review medical records. This is achieved by executing and deploying smart contracts on the blockchain, facilitating a wide range of services.

[0042] The specific requirements can be summarized as follows:

[0043] 1. Following the traditional processing workflow of the electronic medical record system, a series of procedures were developed, including user registration and login, study time management, doctor consultation, patient history query, and patient authorization management.

[0044] 2. The system architecture involves creating a client for three user types: patients, doctors, and administrators. This requires the establishment of their own identity management chain code and its data structure.

[0045] 3. Build a NodeJs server, use the server to call the SDKapi of different chain codes, interact with the system or modify the data stored in the blockchain, and then display the results on the front-end interface.

[0046] like Figure 2 As shown, the system architecture consists of five unique layers: data layer, network layer, consensus layer, contract layer, and application layer. Each layer has a unique role to ensure the seamless functionality of the system. The main data source of the system is electronic medical records (EMRs), which include important patient information such as medical history, diagnosis, treatment plan, and test results. These EMRs form the key to data processing and analysis operations within the system. In order to integrate EMRs into the system, a strict procedure needs to be implemented. EMRs are encapsulated in the blockchain's "block header + block body" format to facilitate efficient storage and retrieval. This format ensures the integrity and security of medical data throughout its life cycle within the system. Before storage, medical data is subject to effective encryption technology.

[0047] The data layer uses advanced hashing algorithms and timestamps to expertly encrypt EMRs. The hashing algorithm converts data into a fixed-size numerical representation, which makes reverse engineering the original content extremely challenging. The timestamp is used to generate a record of when data was added or modified, thereby enhancing the traceability and immutability of the data.

[0048] The network layer plays an integral role in connecting and transmitting encrypted medical data to the blockchain network. It ensures the secure transmission of data between different system components and enables the integration of medical records into the blockchain. By deploying secure communication protocols and encryption techniques, the network layer establishes a reliable and tamper-proof connection, thereby maintaining the confidentiality and integrity of medical data. In summary, the system architecture consists of multiple layers, each serving a specific purpose.

[0049] Example 4: Voice electronic medical record system based on artificial intelligence and blockchain technology

[0050] The electronic medical record system is based on blockchain technology and includes six sub-modules: medical record management module, medical record authority management module, text encryption and decryption module, blockchain client module, wallet management module, and login management module. Each module performs a unique function to help the entire system run smoothly.

[0051] Patients have control over their own electronic medical records, retaining the ability to view and authorize, but lacking the ability to modify or add new records. They need blockchain chain code to write registration or login information. Doctors need to enter their identification number and password to access the system, subsequently accessing the list of registered patients and creating medical records indexed for the patients. Administrators are tasked with managing the list of doctors within their respective departments.

[0052] The interaction between patients, doctors, administrators and blockchain is facilitated by seven chain codes that meet the system requirements. These include department registration and outpatient management, medical record management, patient and doctor information management, etc. Each chain code handles a specific logical operation and can be called by HTTP server routing, ultimately reflecting user-specific operations on the front-end interface.

[0053] This specific embodiment is merely an explanation of the present invention and is not a limitation of the present invention. After reading this specification, those skilled in the art may make non-creative modifications to the present embodiment as needed. However, as long as they are within the scope of the claims of the present invention, they are protected by the patent law.

Claims

1. A voice electronic medical record system based on artificial intelligence and blockchain technology, characterized by: The system includes a medical record management module, a medical record authority management module, a text encryption and decryption module, a blockchain client module, a wallet management module and a login management module.

2. According to claim 1, a voice electronic medical record system based on artificial intelligence and blockchain technology is characterized by: The process of generating electronic medical records in the medical record management module is as follows: (1) Program relevant clinical questions, collect responses through a conversational user interface, extract and summarize keywords and sentences from these responses to populate the electronic medical record, and collect basic patient information, including age, gender, medical history, and other details required for care delivery; (2) Obtain specific consultation information through doctor questions, patient answers, orderly diagnosis and prescription treatment, and then use the Chinese voice network to transcribe the complex voice conversation information between doctors and patients into text data, and use natural language processing (NLP) technology to process these text data and organize them into a structured format; (3) Manual verification and modification by physicians: This feature allows physicians to review and modify automatically generated patient information to ensure the accuracy of the data and to add annotations to imaging reports or physiological indicators from patient examinations.

3. According to claim 1, a voice electronic medical record system based on artificial intelligence and blockchain technology is characterized by: Patients control their own electronic medical records through the medical record rights management module, retaining the ability to view and authorize, but lacking the ability to modify or add new records.

4. According to claim 1, a voice electronic medical record system based on artificial intelligence and blockchain technology is characterized by: The patient writes registration or login information through the login management module; The doctor accesses the system by entering an identification number and password through the login management module, and subsequently accesses a list of registered patients and creates a medical record indexed for the patient; Administrators manage the list of doctors in their respective departments through the login management module.

5. According to claim 1, a voice electronic medical record system based on artificial intelligence and blockchain technology is characterized by: The regional chain client realizes the interaction between patients, doctors, managers and blockchain.

6. The voice electronic medical record system based on artificial intelligence and blockchain technology according to claim 1 is characterized by: The text encryption and decryption module encrypts and decrypts the medical record content to protect the patient's privacy.

7. The voice electronic medical record system based on artificial intelligence and blockchain technology according to claim 1 is characterized by: The wallet management module is used to pay the patient's medical expenses.

Images