Multi-model-based interactive authoring environment for creating shareable medical knowledge

Compressed gastric image generation based on soft-label dataset distillation for medical data sharing

BACKGROUND AND OBJECTIVE

  Sharing of medical data is required to enable the cross-agency flow of healthcare information and construct high-accuracy computer-aided diagnosis systems. However, the large sizes of medical datasets, the massive amount of memory of saved deep convolutional neural network (DCNN) models, and patients^' privacy protection are problems that can lead to inefficient medical data sharing. Therefore, this study proposes a novel soft-label dataset distillation method for medical data sharing.

METHODS

The proposed method distills valid information of medical image data and generates several compressed images with different data distributions for anonymous medical data sharing. Furthermore, our method can extract essential weights of DCNN models to reduce the memory required to save trained models for efficient medical data sharing.

RESULTS

The proposed method can compress tens of thousands of images into several soft-label images and reduce the size of a trained model to a few hundredths of its original size. The compressed images obtained after distillation have been visually anonymized; therefore, they do not contain the private information of the patients. Furthermore, we can realize high-detection performance with a small number of compressed images.

CONCLUSIONS

The experimental results show that the proposed method can improve the efficiency and security of medical data sharing.

Clinical Decision Support System Based on Hybrid Knowledge Modeling: A Case Study of Chronic Kidney Disease-Mineral and Bone Disorder Treatment

Clinical decision support systems (CDSSs) represent the latest technological transformation in healthcare for assisting clinicians in complex decision-making. Several CDSSs are proposed to deal with a range of clinical tasks such as disease diagnosis, prescription management, and medication ordering. Although a small number of CDSSs have focused on treatment selection, areas such as medication selection and dosing selection remained under-researched. In this regard, this study represents one of the first studies in which a CDSS is proposed for clinicians who manage patients with end-stage renal disease undergoing maintenance hemodialysis, almost all of whom have some manifestation of chronic kidney disease-mineral and bone disorder (CKD-MBD). The primary objective of the system is to aid clinicians in dosage prescription by levering medical domain knowledge as well existing practices. The proposed CDSS is evaluated with a real-world hemodialysis patient dataset acquired from Kyung Hee University Hospital, South Korea. Our evaluation demonstrates overall high compliance based on the concordance metric between the proposed CKD-MBD CDSS recommendations and the routine clinical practice. The concordance rate of overall medication dosing selection is 78.27%. Furthermore, the usability aspects of the system are also evaluated through the User Experience Questionnaire method to highlight the appealing aspects of the system for clinicians. The overall user experience dimension scores for pragmatic, hedonic, and attractiveness are 1.53, 1.48, and 1.41, respectively. A service reliability for the Cronbach's alpha coefficient greater than 0.7 is achieved using the proposed system, whereas a dependability coefficient of the value 0.84 reveals a significant effect.

A scoping review of knowledge authoring tools used for developing computerized clinical decision support systems

Objective

Clinical Knowledge Authoring Tools (CKATs) are integral to the computerized Clinical Decision Support (CDS) development life cycle. CKATs enable authors to generate accurate, complete, and reliable digital knowledge artifacts in a relatively efficient and affordable manner. This scoping review aims to compare knowledge authoring tools and derive the common features of CKATs.

Materials and Methods

We performed a keyword-based literature search, followed by a snowball search, to identify peer-reviewed publications describing the development or use of CKATs. We used PubMed and Embase search engines to perform the initial search (n = 1579). After removing duplicate articles, nonrelevant manuscripts, and not peer-reviewed publication, we identified 47 eligible studies describing 33 unique CKATs. The reviewed CKATs were further assessed, and salient characteristics were extracted and grouped as common CKAT features.

Results

Among the identified CKATs, 55% use an open source platform, 70% provide an application programming interface for CDS system integration, and 79% provide features to validate/test the knowledge. The majority of the reviewed CKATs describe the flow of information, offer a graphical user interface for knowledge authors, and provide intellisense coding features (94%, 97%, and 97%, respectively). The composed list of criteria for CKAT included topics such as simulating the clinical setting, validating the knowledge, standardized clinical models and vocabulary, and domain independence. None of the reviewed CKATs met all common criteria.

Conclusion

Our scoping review highlights the key specifications for a CKAT. The CKAT specification proposed in this review can guide CDS authors in developing more targeted CKATs.

Acquiring guideline-enabled data driven clinical knowledge model using formally verified refined knowledge acquisition method

BACKGROUND AND OBJECTIVE

Validation and verification are the critical requirements for the knowledge acquisition method of the clinical decision support system (CDSS). After acquiring the medical knowledge from diverse sources, the rigorous validation and formal verification process are required before creating the final knowledge model. Previously, we have proposed a hybrid knowledge acquisition method with the support of a rigorous validation process for acquiring medical knowledge from clinical practice guidelines (CPGs) and patient data for the treatment of oral cavity cancer. However, due to lack of formal verification process, it involves various inconsistencies in knowledge relevant to the formalism of knowledge, conformance to CPGs, quality of knowledge, and complexities of knowledge acquisition artifacts.

METHODS

This paper presents the refined knowledge acquisition (ReKA) method, which uses the Z formal verification process. The ReKA method adopts the verification method and explores the mechanism of theorem proving using the Z notation. It enhances a hybrid knowledge acquisition method to thwart the inconsistencies using formal verification.

RESULTS

ReKA adds a set of nine additional criteria to be used to have a final valid refined clinical knowledge model. These criteria ensure the validity of the final knowledge model concerning formalism of knowledge, conformance to GPGs, quality of the knowledge, usage of stringent conditions and treatment plans, and inconsistencies possibly resulting from the complexities. Evaluation, using four medical knowledge acquisition scenarios, shows that newly added knowledge in CDSS due to the additional criteria by the ReKA method always produces a valid knowledge model. The final knowledge model was also evaluated with 1229 oral cavity patient cases, which outperformed with an accuracy of 72.57% compared to a similar approach with an accuracy of 69.7%. Furthermore, the ReKA method identified a set of decision paths (about 47.8%) in the existing approach, which results in a final knowledge model with low quality, non-conformed from standard CPGs.

CONCLUSION

ReKA refined the hybrid knowledge acquisition method by discovering the missing steps in the current validation process at the acquisition stage. As a formally proven method, it always yields a valid knowledge model having high quality, supporting local practices, and influenced by standard CPGs. Furthermore, the final knowledge model obtained from ReKA also preserves the performance such as the accuracy of the individual source knowledge models.

Artificial intelligence for the diagnosis of heart failure

The diagnosis of heart failure can be difficult, even for heart failure specialists. Artificial Intelligence-Clinical Decision Support System (AI-CDSS) has the potential to assist physicians in heart failure diagnosis. The aim of this work was to evaluate the diagnostic accuracy of an AI-CDSS for heart failure. AI-CDSS for cardiology was developed with a hybrid (expert-driven and machine-learning-driven) approach of knowledge acquisition to evolve the knowledge base with heart failure diagnosis. A retrospective cohort of 1198 patients with and without heart failure was used for the development of AI-CDSS (training dataset, n = 600) and to test the performance (test dataset, n = 598). A prospective clinical pilot study of 97 patients with dyspnea was used to assess the diagnostic accuracy of AI-CDSS compared with that of non-heart failure specialists. The concordance rate between AI-CDSS and heart failure specialists was evaluated. In retrospective cohort, the concordance rate was 98.3% in the test dataset. The concordance rate for patients with heart failure with reduced ejection fraction, heart failure with mid-range ejection fraction, heart failure with preserved ejection fraction, and no heart failure was 100%, 100%, 99.6%, and 91.7%, respectively. In a prospective pilot study of 97 patients presenting with dyspnea to the outpatient clinic, 44% had heart failure. The concordance rate between AI-CDSS and heart failure specialists was 98%, whereas that between non-heart failure specialists and heart failure specialists was 76%. In conclusion, AI-CDSS showed a high diagnostic accuracy for heart failure. Therefore, AI-CDSS may be useful for the diagnosis of heart failure, especially when heart failure specialists are not available.

Advances in Sharing Multi-sourced Health Data on Decision Support Science 2016-2017

Introduction:  Clinical decision support science is expanding to include integration from broader and more varied data sources, diverse platforms and delivery modalities, and is responding to emerging regulatory guidelines and increased interest from industry.

Objective:  Evaluate key advances and challenges of accessing, sharing, and managing data from multiple sources for development and implementation of Clinical Decision Support (CDS) systems in 2016-2017.

Methods:  Assessment of literature and scientific conference proceedings, current and pending policy development, and review of commercial applications nationally and internationally.

Results:  CDS research is approaching multiple landmark points driven by commercialization interests, emerging regulatory policy, and increased public awareness. However, the availability of patient-related “Big Data” sources from genomics and mobile health, expanded privacy considerations, applications of service-based computational techniques and tools, the emergence of “app” ecosystems, and evolving patient-centric approaches reflect the distributed, complex, and uneven maturity of the CDS landscape. Nonetheless, the field of CDS is yet to mature. The lack of standards and CDS-specific policies from regulatory bodies that address the privacy and safety concerns of data and knowledge sharing to support CDS development may continue to slow down the broad CDS adoption within and across institutions.

Conclusion:  Partnerships with Electronic Health Record and commercial CDS vendors, policy makers, standards development agencies, clinicians, and patients are needed to see CDS deployed in the evolving learning health system.