Information technology outside health care: what does it matter to us?

Non-health-care uses of information technology (IT) provide important lessons for health care informatics that are often overlooked because of the focus on the ways in which health care is different from other domains. Eight examples of IT use outside health care provide a context in which to examine the content and potential relevance of these lessons. Drawn from personal experience, five books, and two interviews, the examples deal with the role of leadership, academia, the private sector, the government, and individuals working in large organizations. The interviews focus on the need to manage technologic change. The lessons shed light on how to manage complexity, create and deploy standards, empower individuals, and overcome the occasional "wrongness" of conventional wisdom. One conclusion is that any health care informatics self-examination should be outward-looking and focus on the role of health care IT in the larger context of the evolving uses of IT in all domains.

Potato-lactulose breath hydrogen testing as a function of gastric motility in diabetes mellitus.

OBJECTIVE: Scintigraphic determination of gastric emptying is the current standard for the assessment of gastric motility and the diagnosis of diabetic gastroparesis. However, such studies are expensive, inconvenient, and involve exposure to radiation. Because the time course of breath hydrogen (H2) excretion after ingestion of lactulose correlates with upper gastrointestinal transit time, we hypothesized that patients with diabetic gastroparesis would exhibit prolonged breath H2 excretion after ingestion of a test meal containing complex carbohydrate and lactulose compared to subjects without diabetes and subjects with diabetes but without gastroparesis. RESEARCH DESIGN AND METHODS: Ten healthy subjects without diabetes, 10 subjects with diabetes but without gastroparesis (gastric emptying T1/2,T1/2 < 90 minutes), and 10 subjects with diabetes and previously diagnosed gastroparesis (T1/2 > 90 minutes) were admitted for a single 24-hour study. Gastric motility agents were withheld 24 hours prior to the study. Euglycemia was established and maintained overnight in subjects with diabetes with continuous intravenous insulin infusion. At 6:00 AM, all subjects ingested a breakfast containing 100 g of cooked potato starch and 20 g lactulose. Breath H2 excretion was monitored at baseline and every 30 minutes for 12 hours after ingestion of the test meal. RESULTS: Twelve hours after ingestion of the test meal, raw and baseline adjusted breath H2 excretion was significantly elevated in the gastroparesis group compared to the unaffected group with diabetes and the group without diabetes (p < 0.001). The baseline and 12-hour data points were adequate to discriminate between normal and delayed gastric emptying. CONCLUSIONS: We conclude that patients with previously diagnosed gastroparesis exhibit prolonged breath H2 excretion after ingestion of a test meal. This test may prove to be a safe, reliable, and affordable outpatient screening test for diabetic gastroparesis.

A poor precedent.

The single greatest impediment to the successful leveraging of computer-based patient records (CPRs) is the difficulty of creating and maintaining comparable patient descriptions. Specifically, it will be hard to justify the investment required to deploy CPRs widely if the patient descriptions they store are not comparable across successive releases of controlled health-care vocabularies. Thus, it is necessary to solve the controlled health-care vocabulary update problem for CPRs before the comparability of patient descriptions can be sustained. What may seem to be a narrow technical problem of interest only to maintainers of health-care enterprise systems is, instead, a central problem of medical informatics. Cimino's "Formal descriptions and adaptive mechanisms for changes in controlled medical vocabularies" describes a classification of the changes appearing in the 1994 Edition of the International Classification of Diseases (ICD-9-CM). His paper describes the conversion of differences detected between the 1993 and 1994 releases of ICD-9-CM and a conversion of the elements of the classification into the required formal changes to the Medical Entities Dictionary (MED), part of the CPR in use at Columbia Presbyterian Medical Center. Because the process of detecting differences begins with an empirical analysis of the ASCII representations of the 1993 and 1994 releases of ICD-9-CM, it is impossible for a computer program to infer the intent of the changes that caused the differences; instead, a content expert must infer the intent and then update the MED accordingly. A typical task is to infer whether a change in naming also reflects a change in the meaning named. While Cimino's methods and their execution are exemplary in every respect, they nevertheless constitute a kind of "reverse engineering"-an ad hoc attempt to infer intent from details. Reverse engineering of changes to controlled medical vocabularies is a poor precedent. Such methods should be viewed as necessary short-term expedients only, and all parties concerned should work toward an incremental plan by which the intent of changes to controlled health-care vocabularies can be made both explicit and machine processible. Only then can the comparability of patient descriptions be sustained.

Navigating to knowledge.

One way to fulfill point-of-care knowledge needs is to present caregivers with a visual representation of the available "answers". Using such a representation, caregivers can recognize what they want, rather than have to recall what they need, and then navigate to an appropriate answer. Given selected pieces of information from a computer-based patient record, an interface can anticipate certain knowledge needs by initializing caregiver navigation in a semantic neighborhood of answers likely to be relevant to the patient at hand. These notions draw heavily on two collaborative projects--the U.S. National Library of Medicine Unified Medical Language System and the U.S. National Cancer Institute Knowledge Server. Both of these projects support navigation because they make the structure of medical knowledge explicit in a way that can be exploited by human interfaces.

Metaphrase: an aid to the clinical conceptualization and formalization of patient problems in healthcare enterprises.

Patient descriptors, or "problems," such as "brain metastases of melanoma" are an effective way for caregivers to describe patients. But most problems, e.g., "cubital tunnel syndrome" or "ulnar nerve compression," found in problem lists in an Electronic Medical Record (EMR) are not comparable computationally--in general, a computer cannot determine whether they describe the same or a related problem, or whether the user would have preferred "ulnar nerve compression syndrome." Metaphrase is a scalable, middleware component designed to be accessed from problem-manager applications in EMR systems. In response to caregivers' informal descriptors it suggests potentially equivalent, authoritative, and more formally comparable descriptors. Metaphrase contains a clinical subset of the 1997 UMLS Metathesaurus and some 10,000 "problems" from the Mayo Clinic and Harvard Beth Israel Hospital. Word and term completion, spelling correction, and semantic navigation, all combine to ease the burden of problem conceptualization, entry and formalization.

The role of the UMLS in 'storing' and 'sharing' across systems.

We will argue that 'sharing', 're-use', 're-purposing', and 'addition' of health care information is difficult, intrinsically; that the best way to overcome the difficulty is to start doing it, as soon as possible, and that the UMLS Knowledge Sources provide the best place to start. We recommend that the UMLS be used as a default source of biomedical concept names and relationships, as a comprehensive, data-based, 'reference model', and as an example of a large, ecumenical, evolving, continuously updated source of re-usable health care information.

A "lexically-suggested logical closure" metric for medical terminology maturity.

Medical Terminologies are becoming increasingly expressive secondary to their increase in size, and are becoming increasingly difficult to analyze secondary to inconsistencies in their use and complex interrelationships that are often not explicitly defined. To address these problems, SNOMED-RT is being developed to allow consistent use, and to define explicitly interrelationships between terms. Ensuring the quality of a terminology system like SNOMED-RT presents new challenges which we are trying to address with theoretically-grounded methodologies for quality management. Here we describe an initial metric toward achieving this goal called "lexically-suggested logical closure." We explain how this metric can be useful for tracking the maturity and quality of a terminology, and apply this metric to track the progress of SNOMED-RT development over a portion of its life-cycle.

The UMLS Metathesaurus: representing different views of biomedical concepts.

The UMLS Metathesaurus is a compilation of names, relationships, and associated information from a variety of biomedical naming systems representing different views of biomedical practice or research. The Metathesaurus is organized by meaning, and the fundamental unit in the Metathesaurus is the concept. Differing names for a biomedical meaning are linked in a single Metathesaurus concept. Extensive additional information describing semantic characteristics, occurrence in machine-readable information sources, and how concepts co-occur in these sources is also provided, enabling a greater comprehension of the concept in its various contexts. The Metathesaurus is not a standardized vocabulary; it is a tool for maximizing the usefulness of existing vocabularies. It serves as a knowledge source for developers of biomedical information applications and as a powerful resource for biomedical information specialists.

Desiderata for a clinical terminology server.

Clinical terminology servers are distinguished from more broadly based terminology servers intended for nomenclature development or mediation across classifications. Focusing upon the consistent and comparable entry of clinical observations, findings, and events, key desiderata are enumerated and expanded. These include 1) word normalization, 2) word completion, 3) target terminology specification, 4) spelling correction, 5) lexical matching, 6) term completion, 7) semantic locality, 8) term composition and 9) decomposition. Comparisons of this functionality to previously published models and specifications are made. Experience with a clinical terminology server, Metaphrase, is described.

Integrating pharmacokinetics knowledge into a drug ontology: as an extension to support pharmacogenomics.

The newly developed U.S. Common Medication Information Infrastructure was used as a basis to capture and formally express the properties of drugs relevant to research and the clinical application of pharmacogenomics. Two associated taxonomies within the model, Mechanism of Action and Physiologic Effect, were enriched to accommodate pharmacogenomic use-cases; the 4,000 active ingredients in the VA NDF-RT drug file were related to the enhanced taxonomies. Pharmacokinetics were independently modeled for pharmacogenomics and tested against thirty-one high-profile drugs to demonstrate our approach.

The semantic structure of the UMLS Metathesaurus.

Meta-1.1, the UMLS metathesaurus, represents medical knowledge in the forms of names of concepts and links between those concepts. The representations of the semantic neighborhood of a concept can be thought of as dimensions of the property of semantic locality and include term information (broader, narrower, or otherwise related), the contextual information (parent-child, siblings in a hierarchy), the semantic types, and the co-occurrence data (links discovered empirically from concepts used to index the medical literature.) The degree of redundancy of each of these dimensions was investigated by reviewing the extent of multiple presentations of concepts which appear as related to a given concept. The degree of overlap was surprisingly small. While the co-occurrence data finds some of the links represented by other dimensions, those links are but minute fractions of the vast amount of co-occurrence derived links. Because parent-child relationships are often subsumptive (or categorical) in nature, it might be expected that siblings usually share the same semantic types. While true in the aggregate, the wide variance in percent of types shared may reflect the intended usages of the source vocabularies. Noun phrases were extracted from the definitions of 40 concepts in Meta-1 in order to assess systematically the coverage of important concepts by Meta-1, and to assess whether the links between these definitional concepts, which may have special value, and the concept being defined were indeed present. Out of 161 of these definitional concepts, 29 were not represented in Meta-1, and 37 of those represented in Meta-1 had no direct link to the concept they were defining.(ABSTRACT TRUNCATED AT 250 WORDS)

Recognizing new medical knowledge computationally.

Can new medical knowledge be recognized computationally? We know knowledge is changing, and our knowledge-based systems will need to accommodate that change in knowledge on a regular basis if they are to stay successful. Computational recognition of these changes seems desirable. It is unlikely that low level objects in the computational universe, bits and characters, will change much over time, higher level objects of language, where meaning begins to emerge, may show change. An analysis of ten arbitrarily selected paragraphs from the Medical Knowledge Self-Assessment Program of the American College of Physicians was used as a test bed for nominal phrase recognition. While there were words not known to Meta-1.2, only 8 of the 32 concepts new to the primary author were pointed to by new words. Use of a barrier word method was successful in identifying 23 of the 32 new concepts. Use of co-occurrence (in sentences) of putative nominal phrases may reduce the amount of human effort involved in recognizing the emergence of new relationships.

Categorization of free-text problem lists: an effective method of capturing clinical data.

Problem lists assist in organizing patient information in computer based medical records. However, in order to use problem lists for billing, research, decision support and standardization, a categorization of the problems entered is required. We describe the problem list component of our computerized patient record, the On-line Medical Record (OMR), which combines a free-text entry mechanism with a categorization scheme, using a dictionary containing 846 terms. All 118,040 problems entered during the system's six years of use have been analyzed, 477 clinicians have entered a mean +/- S.D. of 238 +/- 604 problems into 22,311 patient records. The average number of problems in each patient's file was 5.1 +/- 3.9. Comments were typed for 80,281 (68%) of the problems, ranging in length from 1 to 2456 characters, with a mean length of 98 +/- 110 characters. Half the problems were entered on the day of the encounter with the patient. Overall, 66% of all problems were categorized in relation to terms from the problem dictionary. Lexical analysis of all problem names showed that 80% could be mapped to Meta 1.4, Snomed 3.0 or a pre-release version of Read 3.0. We conclude that a problem list entry scheme combining free-text entry and optional categorization using a dictionary can result in a high proportion of problems being categorized as desired. Improvement of the system by elimination of unused dictionary terms and addition of 1000 terms identified by the lexical analysis is likely to result in even higher categorization rates.

Semantic visualization of oncology knowledge sources.

Visualization of knowledge sources can have a substantial impact on the use of such sources at the point of care. This is because barriers to use at the point of care include hours required to master the electronic interfaces to those sources, and minutes required to master the electronic interfaces to those sources, and minutes required to accomplish any one retrieval. For a system to be used regularly at the point of care, therefore, it must be intuitive and fast. This paper presents a three dimensional interface to oncology knowledge sources that aims to meet this challenge.

Identifying concepts in medical knowledge.

The barrier word method of identifying nominal phrases in text, using a very long barrier word list, was evaluated in two different sets of text. In a sample of 10 paragraphs from the Medical Knowledge Self-Assessment Program of the American College of Physicians, the yield of nominal phrases as a percent of total chunks isolated was 66%. Some 500,000 chunks were isolated from Principles and Practice of Oncology (PPO). 38% of these chunk-occurrences were of chunks which matched to 10,000 concept names in Meta-1.4, the most recent version of the UMLS Metathesaurus. 50 paragraphs from PPO were chosen at random. Co-occurrences of concepts in those paragraphs were reviewed. 42 of the paragraphs had unique or infrequently occurring co-occurrences which described closely the major thrust of the paragraph.

Judging prognosis in malignant melanoma of the skin. A problem of inference over small data sets.

Data was generated for 828 clinical stage 1 melanoma patients, divided into groups according to sex, tumor location, and tumor level for each of the 56 groups. Summary data, including the number of patients, number of patients dying as a result of melanoma, range of tumor thickness, mean and median tumor thickness, and mean length of follow-up of the surviving patients, are shown. Patients with melanoma of the palms and soles, subungual melanomas, and mucosal melanomas were excluded. A physician with a new melanoma patient could select the appropriate group for his or her patient, matched with respect to sex, location, and level, and then make a judgment regarding the prognosis, based on the survival experience of the group. In a few groups, the small numbers of patients provides only a rough impression of survival, but with many groups, a fair estimate can be made. The effectiveness of elective lymph node dissection was examined by creating 111 pairs of patients, matched by sex, level, location, and tumor thickness (to within +/- 12%), in which one member of the pair had an elective node dissection (ELND) and the other did not. There was no statistically significant difference between the survival of the two groups.

Malignant melanoma of the skin. I. The association of tumor depth and type, and patient sex, age, and site with survival.

The occurrence and behavior of cutaneous melanomas in a group of 1123 patients studied prospectively, is described in terms of histologic type, tumor thickness and levels of invasion, the patients' sex and age, and the anatomic location of the primary tumors. Associations amongst these attributes, and with survival, are also examined. The characteristics of the patients in this study (who on average are somewhat younger, and have better prognoses and survivals than those reported by most other groups) are compared with data obtained (primarily over the past decade) in other geographical areas, and with different patient populations. Evidence is presented that sex, tumor location, and age (in the case of males) are also predictive of survival.