Watson for Oncology and breast cancer treatment recommendations: agreement with an expert multidisciplinary tumor board.

Background: Breast cancer oncologists are challenged to personalize care with rapidly changing scientific evidence, drug approvals, and treatment guidelines. Artificial intelligence (AI) clinical decision-support systems (CDSSs) have the potential to help address this challenge. We report here the results of examining the level of agreement (concordance) between treatment recommendations made by the AI CDSS Watson for Oncology (WFO) and a multidisciplinary tumor board for breast cancer. Patients and methods: Treatment recommendations were provided for 638 breast cancers between 2014 and 2016 at the Manipal Comprehensive Cancer Center, Bengaluru, India. WFO provided treatment recommendations for the identical cases in 2016. A blinded second review was carried out by the center's tumor board in 2016 for all cases in which there was not agreement, to account for treatments and guidelines not available before 2016. Treatment recommendations were considered concordant if the tumor board recommendations were designated 'recommended' or 'for consideration' by WFO. Results: Treatment concordance between WFO and the multidisciplinary tumor board occurred in 93% of breast cancer cases. Subgroup analysis found that patients with stage I or IV disease were less likely to be concordant than patients with stage II or III disease. Increasing age was found to have a major impact on concordance. Concordance declined significantly (P ≤ 0.02; P < 0.001) in all age groups compared with patients <45 years of age, except for the age group 55-64 years. Receptor status was not found to affect concordance. Conclusion: Treatment recommendations made by WFO and the tumor board were highly concordant for breast cancer cases examined. Breast cancer stage and patient age had significant influence on concordance, while receptor status alone did not. This study demonstrates that the AI clinical decision-support system WFO may be a helpful tool for breast cancer treatment decision making, especially at centers where expert breast cancer resources are limited.

Commentaries on the IMIA Award of Excellence Lecture by Reinhold Haux.

The IMIA Yearbook editorial team asked five internationally renowned biomedical informaticians to respond to Prof. Haux's editorial. This paper summarizes their thoughts and responses. Contributions are ordered alphabetically by the contributor's last name. All authors provided an equal contribution to this manuscript.

Lessons learned from a health record bank start-up.

INTRODUCTION: This article is part of a Focus Theme of METHODS of Information in Medicine on Health Record Banking. BACKGROUND: In late summer 2010, an organization was formed in greater Phoenix, Arizona (USA), to introduce a health record bank (HRB) in that community. The effort was initiated after market research and was aimed at engaging 200,000 individuals as members in the first year (5% of the population). It was also intended to evaluate a business model that was based on early adoption by consumers and physicians followed by additional revenue streams related to incremental services and secondary uses of clinical data, always with specific permission from individual members, each of whom controlled all access to his or her own data. OBJECTIVES: To report on the details of the HRB experience in Phoenix, to describe the sources of problems that were experienced, and to identify lessons that need to be considered in future HRB ventures. METHODS: We describe staffing for the HRB effort, the computational platform that was developed, the approach to marketing, the engagement of practicing physicians, and the governance model that was developed to guide the HRB design and implementation. RESULTS: Despite efforts to engage the physician community, limited consumer advertising, and a carefully considered financial strategy, the experiment failed due to insufficient enrollment of individual members. It was discontinued in April 2011. CONCLUSIONS: Although the major problem with this HRB project was undercapitalization, we believe this effort demonstrated that basic HRB accounts should be free for members and that physician engagement and participation are key elements in constructing an effective marketing channel. Local community governance is essential for trust, and the included population must be large enough to provide sufficient revenues to sustain the resource in the long term.

50 years of informatics research on decision support: what's next.

OBJECTIVES: To reflect on the history, status, and future trends of decision support in health and biomedical informatics. To highlight the new challenges posed by the complexity and diversity of genomic and clinical domains. To examine the emerging paradigms for supporting cost-effective, personalized decision making. METHODS: A group of international experts in health and biomedical informatics presented their views and discussed the challenges and issues on decision support at the Methods of Information in Medicine 50th anniversary symposium. The experts were invited to write short articles summarizing their thoughts and positions after the symposium. RESULTS AND CONCLUSIONS: The challenges posed by the complexity and diversity of the domain knowledge, system infrastructure, and usage pattern are highlighted. New requirements and computational paradigms for representing, using, and acquiring biomedical knowledge and healthcare protocols are proposed. The underlying common themes identified for developing next-generation decision support include incorporating lessons from history, uniform vocabularies, integrative interfaces, contextualized decisions, personalized recommendations, and adaptive solutions.

Biomedical informatics in the desert--a new and unique program at Arizona State University.

OBJECTIVES: A new academic Biomedical Informatics (BMI) Program in Phoenix, Arizona, embodies a unique organizational structure to draw on the strengths of a computer science and informatics school and the biomedical and clinical strengths of a college of medicine, in an effort to infuse informatics approaches broadly. METHODS: The program reflects a partnership of two state universities that situates the Arizona State University (ASU) Department of BMI on a new downtown Phoenix Biomedical Campus with the University of Arizona (UA) College of Medicine in partnership with ASU (COM-PHX). Plans call for development of faculty and expertise in the four major subdomains of BMI, as well as in various cross-cutting capabilities. RESULTS: Coming into existence in a state that is investing significantly in biomedical science and technology, BMI has already developed Masters and PhD degree programs, is working with COM-PHX to integrate informatics intensively into the education of the medical students, and has been authorized to plan for an undergraduate program in BMI. Reflecting the statewide emphasis on the biomedical and health sector, the growing faculty are engaged in a number of research partnerships and collaborative centers. CONCLUSIONS: As one of the newest academic BMI programs is taking shape in Arizona, it is embarking on a wide-ranging educational program and a broad research agenda that are now in their earliest stages.

On exemplary scientific conduct regarding submission of manuscripts to biomedical informatics journals.

As the Editors of leading international biomedical informatics journals, the authors report on a recent pattern of improper manuscript submissions to journals in our field. As a guide for future authors, we describe ethical and pragmatic issues related to submitting work for peer-reviewed journal publication. We propose a coordinated approach to the problem that our respective journals will follow. This Editorial is being jointly published in the following journals represented by the authors: Computer Methods and Programs in Biomedicine, International Journal of Medical Informatics, Journal of Biomedical Informatics, Journal of the American Medical Informatics Association, and Methods of Information in Medicine.

Toward a representation format for sharable clinical guidelines.

Clinical guidelines are being developed for the purpose of reducing medical errors and unjustified variations in medical practice, and for basing medical practice on evidence. Encoding guidelines in a computer-interpretable format and integrating them with the electronic medical record can enable delivery of patient-specific recommendations when and where needed. Since great effort must be expended in developing high-quality guidelines, and in making them computer-interpretable, it is highly desirable to be able to share computer-interpretable guidelines (CIGs) among institutions. Adoption of a common format for representing CIGs is one approach to sharing. Factors that need to be considered in creating a format for sharable CIGs include (i) the scope of guidelines and their intended applications, (ii) the method of delivery of the recommendations, and (iii) the environment, consisting of the practice setting and the information system in which the guidelines will be applied. Several investigators have proposed solutions that improve the sharability of CIGs and, more generally, of medical knowledge. These approaches can be useful in the development of a format for sharable CIGs. Challenges in sharing CIGs also include the need to extend the traditional framework for disseminating guidelines to enable them to be integrated into practice. These extensions include processes for (i) local adaptation of recommendations encoded in shared generic guidelines and (ii) integration of guidelines into the institutional information systems.

Sharable representation of clinical guidelines in GLIF: relationship to the Arden Syntax.

Clinical guidelines are intended to improve the quality and cost effectiveness of patient care. Integration of guidelines into electronic medical records and order-entry systems, in a way that enables delivery of patient-specific advice at the point of care, is likely to encourage guidelines acceptance and effectiveness. Among the methodologies for modeling guidelines and medical decision rules, the Arden Syntax for Medical Logic Modules and the GuideLine Interchange Format version 3 (GLIF3) emphasize the importance of sharing encoded logic across different medical institutions and implementation platforms. These two methodologies have similarities and differences; in this paper we clarify their roles. Both methods can be used to support sharing of medical knowledge, but they do so in complementary situations. The Arden Syntax is suitable for representing individual decision rules in self-contained units called Medical Logic Modules (MLMs), which are usually implemented as event-driven alerts or reminders. In contrast, GLIF3 is designed for encoding complex multistep guidelines that unfold over time. As a consequence, GLIF3 has several mechanisms for complexity management and additional constructs that may require overhead unnecessary for expressing simple alerts and reminders. Unlike the Arden Syntax, GLIF3 encourages a top-down process of guideline modeling consisting of three levels that are created in order: Level 1 comprises a human-readable flowchart of clinical decisions and actions. Level 2 comprises a computable specification that can be verified for logical consistency and completeness; and Level 3 comprises an implementable specification that includes information required for local adaptation of guideline logic as well as for mapping guideline variables onto institutional medical records. A major emphasis of the current GLIF3 development process has been to create the computable specification that formally represents medical decision and eligibility criteria. We based GLIF3's formal expression language on the Arden Syntax's logic grammar, making the necessary extensions to the Arden Syntax's data structures and operators to support GLIF3's object-oriented data model. We discuss why the process of generating a set of MLMs from a GLIF-encoded guideline cannot be automated, why it can result in information loss, and why simple medical rules are best represented as individual MLMs. We thus show that the Arden Syntax and GLIF3 play complementary roles in representing medical knowledge for clinical decision support.

Issues and opportunities in public health informatics: a panel discussion.

A panel was convened at the American Medical Informatics Association Spring Congress to discuss issues and opportunities that arise when informatics methods, theories, and applications are applied to public health functions. Panelists provided examples of applications that connect efforts between public health and clinical care, emphasizing the need for integration of clinical data with public health data and the analysis of those data to support surveillance and informed decision making. Benefits to be gained by both medical informatics and public health at the interface were evident; both encounter the same major issues including privacy, systems integration, standards, and many more.

Sharable computer-based clinical practice guidelines: rationale, obstacles, approaches, and prospects.

Clinical practice guideline automation at the point of care is of growing interest, yet most guidelines are authored in unstructured narrative form. Computer-based execution depends on a formal structured representation, and also faces a number of other challenges at all stages of the guideline lifecycle: modeling, authoring, dissemination, implementation, and update. This is because of the multiplicity of conceptual models, authoring tools, authoring approaches, intended applications, implementation platforms, and local interface requirements and operational constraints. Complexity and time required for development and structure are also huge obstacles. These factors argue for convergence on a common shared model for representation that can be the basis of dissemination. A common model would facilitate direct interpretation or mapping to multiple implementation environments. GLIF (GuideLine Interchange Format) is a formal representation model for guidelines, created by the InterMed Collaboratory as a proposed basis for a shared representation. GLIF currently addresses the process of authoring and dissemination; the InterMed team's major focus now is on tools to facilitate these tasks and the mapping to clinical information system environments. Because of limitations in what can be done by a single team with finite resources, however, and the variety of additional perspectives that need to be accommodated, the InterMed team has determined that further development of a shared representation would be best served as an open process in which the world community is engaged. Under the auspices of the HL7 Decision Support Technical Committee, a GLIF Special Interest Group has been established, which is intended to be a forum for collaborative refinement and extension of a standard representation that can support the needs of the guideline lifecycle. Significant areas for future work will need to include demonstrations of effective means for incorporating guide-lines at point of care, reconciliation of functional requirements of different models and identification of those most important for supporting practical implementation, im-proved means for authoring and management of complexity, and methods for automatically analyzing and validating syntax, semantics, and logical consistency of guidelines.

Handling expressiveness and comprehensibility requirements in GLIF3.

Clinical guidelines are aimed at standardizing patient care and improving its quality and cost effectiveness. Guidelines represented in a computer-interpretable (CI) format can be used to provide automatic decision support applied to individual patients during the clinical encounter. The process of creating computer-interpretable guidelines (CIG) re-moves ambiguities contained in paper-based guidelines, thus making the guideline more comprehensible. For these reasons, CIGs may have a larger impact on clinician behavior than paper-based guidelines. Since much effort goes into creating guidelines in a CI format, it is desirable that different institutions and software systems share them. In a guideline representation workshop hosted by the InterMed Collaboratory in March 2000, the need for a standard representation format for sharable CIGs was recognized. As a first step towards achieving this goal, we proposed a set of functional requirements for sharable CIGs. The requirements encompass the entire life cycle of a CIG: development, implementation, use and maintenance. In this paper we discuss requirements that are important during the development stage of a CIG. We have abstracted the requirements into two groups: expressiveness--the ability to ex-press the knowledge content of different types of guidelines--and comprehensibility--the ability to manage complexity, facilitate coherence, and visualize a guideline model to aid in human comprehension. The Guideline Interchange For-mat version 3 (GLIF3) is a language for structured representation of CIGs. It is under development to facilitate sharing CIGs among different institutions and systems. We illustrate how GLIF3 meets the specified development requirements.

Representation of clinical practice guidelines for computer-based implementations.

Representation of clinical practice guidelines is a critical issue for computer-based guideline development, implementation and evaluation. We studied eight types of computer-based guideline representation models. Typical primitives for these models include decisions, actions, patient states and execution states. Temporal constraints and nesting are important aspects of guideline structure representation. Integration of guidelines with electronic medical records can be facilitated by the introduction of formal models of patient data. Patient states and execution states are closely related to one another. Data collection, decision, patient state and intervention are four basic steps in a guideline's logic flow.

Methods of cognitive analysis to support the design and evaluation of biomedical systems: the case of clinical practice guidelines.

This article provides a theoretical and methodological framework for the use of cognitive analysis to support the representation of biomedical knowledge and the design of clinical systems, using clinical-practice guidelines (CPGs) as an example. We propose that propositional and semantic analyses, when used as part of the system-development process, can improve the validity, usability, and comprehension of the resulting biomedical applications. The framework we propose is based on a large body of research on the study of how people mentally represent information and subsequently use it for problem solving. This research encompasses many areas of psychology, but the more important ones are the study of memory and the study of comprehension. Of particular relevance is research devoted to investigating the comprehension and memory of language, expressed verbally or in text. In addition, research on how contextual variables affect performance is informative because these psychological processes are influenced by situational variables (e.g., setting, culture). One important factor limiting the acceptance and use of clinical-practice guidelines (CPGs) may be the mismatch between a guideline's recommended actions and the physician-user's mental models of what seems appropriate in a given case. Furthermore, CPGs can be semantically complex, often composed of elaborate collections of prescribed procedures with logical gaps or contradictions that can promote ambiguity and hence frustration on the part of those who attempt to use them. An improved understanding of the semantics and structure of CPGs may help to improve such matching, and ultimately the comprehensibility and usability of CPGs. Cognitive methods of analysis can help guideline designers and system builders throughout the development process, from the conceptual design of a computer-based system to its implementation phases. By studying how guideline creators and developers represent guidelines, both mentally and in text, and how end-users understand and make decisions with such guidelines, we can inform the development of technologies that seek to improve the match between the representations of experts and practitioners. We urge informaticians to recognize the potential relevance of cognitive analysis methods and to begin more extensive experimentation with the their use in biomedical informatics research.

Using features of Arden Syntax with object-oriented medical data models for guideline modeling.

Computer-interpretable guidelines (CIGs) can deliver patient-specific decision support at the point of care. CIGs base their recommendations on eligibility and decision criteria that relate medical concepts to patient data. CIG models use expression languages for specifying these criteria, and define models for medical data to which the expressions can refer. In developing version 3 of the GuideLine Interchange Format (GLIF3), we used existing standards as the medical data model and expression language. We investigated the object-oriented HL7 Reference Information Model (RIM) as a default data model. We developed an expression language, called GEL, based on Arden Syntax's logic grammar. Together with other GLIF constructs, GEL reconciles incompatibilities between the data models of Arden Syntax and the HL7 RIM. These incompatibilities include Arden's lack of support for complex data types and time intervals, and the mismatch between Arden's single primary time and multiple time attributes of the HL7 RIM.

Guideline classification to assist modeling, authoring, implementation and retrieval.

The National Guideline Clearinghouse (NGC) and its guideline classification system are significant contributions to the study of clinical practice guidelines (CPGs) and their incorporation into routine clinical care. The NGC classification system is primarily designed to support guideline retrieval. We believe that a guideline classification system should also support identification of features that relate to incorporation of executable CPGs into computer-based applications for sharing and delivering guideline-based advice. We have developed a proposed expansion of the NGC guideline classification for this purpose. The axes of the proposed scheme have implications for designing formal models and structures for representing and authoring CPGs. This scheme also has implications for future research.

GLIF3: the evolution of a guideline representation format.

The Guideline Interchange Format (GLIF) is a language for structured representation of guidelines. It was developed to facilitate sharing clinical guidelines. GLIF version 2 enabled modeling a guideline as a flowchart of structured steps, representing clinical actions and decisions. However, the attributes of structured constructs were defined as text strings that could not be parsed, and such guidelines could not be used for computer-based execution that requires automatic inference. GLIF3 is a new version of GLIF designed to support computer-based execution. GLIF3 builds upon the framework set by GLIF2 but augments it by introducing several new constructs and extending GLIF2 constructs to allow a more formal definition of decision criteria, action specifications and patient data. GLIF3 enables guideline encoding at three levels: a conceptual flowchart, a computable specification that can be verified for logical consistency and completeness, and an implementable specification that can be incorporated into particular institutional information systems.

Are medical informatics and nursing informatics distinct disciplines? The 1999 ACMI debate.

The 1999 debate of the American College of Medical Informatics focused on the proposition that medical informatics and nursing informatics are distinctive disciplines that require their own core curricula, training programs, and professional identities. Proponents of this position emphasized that informatics training, technology applications, and professional identities are closely tied to the activities of the health professionals they serve and that, as nursing and medicine differ, so do the corresponding efforts in information science and technology. Opponents of the proposition asserted that informatics is built on a re-usable and widely applicable set of methods that are common to all health science disciplines, and that "medical informatics" continues to be a useful name for a composite core discipline that should be studied by all students, regardless of their health profession orientation.