Search

This paper deals with a major difficulty and potential limiting factor in present-day decision support-that of assigning precise value to an item (or group of items) of clinical information. Historical determinist descriptive thinking has been challenged by current concepts of uncertainty and probability, but neither view is adequate. Four equations are proposed outlining factors which affect the value of clinical information, which explain some previously puzzling observations concerning decision support. It is suggested that without accommodation of these concepts, computer-aided decision support cannot progress further, but if they can be accommodated in future programs, the implications may be profound.

This presentation describes the use of computer aided decision support in acute abdominal pain. The development of such support and the feasability of providing it are described with reference to worldwide studies involving nearly 100,000 patients in the UK, the European Community and worldwide. This presentation will give an overview of the experienced gained in these and other studies. As a result of this experience, it will be suggested: a) The diagnosis of acute abdominal pain by inexperienced emergency surgeons remains a difficult problem. 2) Doctors who have been assisted by a computer have, in many instances, improved their diagnostic and decision-making performance. 3) This improvement has not been due to superior "artificial intelligence" of the computer--but due to the computer acting as an educational focus and a stimulus to good clinical practice. 4) On an international level, these studies have been immensely valuable. They have helped bring together different national groups and helped to develop common medical terminology--as well as foster collaborative "spin-off" in terms of research around the world.

This chapter explores some of the prospects for and problems of computer-aided decision support in clinical medicine, with special reference to acute abdominal pain of both surgical and gynaecological origin. The need for such systems is argued, and it is further demonstrated that their provision is both feasible and (where implemented) associated with tangible improvements in performance by inexperienced hospital doctors. The problems of implementation are nevertheless formidable and these are discussed.

This presentation describes the use of computer aided decision support in acute abdominal pain. The need for such support is explored and the feasability of providing support is described with reference to studies involving nearly 100,000 patients. It is argued that the provision of computer aided decision support can lead to substantial and practical benefit in clinical care--but this is mostly due to the constant stimulus towards "doing it right". This in turn depends upon the provision of a consensus view of "good medicine" in the area concerned; and on an international level strongly argues the case for multi-national cooperative studies to define good medicine and make it available. One such study (the European Community Concerted Action on Acute Abdominal Pain) is described.

This paper describes experience in a modern district general hospital with a small desktop system for computer-aided diagnosis of acute abdominal pain, over a 12-year period involving 5512 cases. When compared with a baseline year (1973) in which unaided performance was monitored, during an initial study period (1974-76) the diagnostic accuracy of junior staff rose by between 10 and 15%. This higher performance level was then maintained for a decade (1976-86) despite changes in staff. The perforation rate among appendicitis cases fell from 27% to 12.5%, accompanied by a smaller fall in negative laparotomy rates. The saving in surgical bednights devoted to acute abdominal pain was approximately 15%, and the notional cost of resources saved during the first 6 years of operation was 120,000 pounds. Other hospitals have shown--in the short term--benefits similar to those obtained at Airedale District General Hospital. The long-term benefits of the system at Airedale reinforce the conclusions of the earlier short-term trials that a comparable system should probably be offered to all DGHs in the UK, not as an exercise in 'artificial intelligence' but as an effective continuing stimulus to good clinical practice.

To assess the diagnostic accuracy of a computer-aided-diagnosis system when implemented in different parts of the world, an automated system, which had established its reliability in Leeds, England, was transferred to Sherbrooke, Quebec. In this preliminary study two retrospective series, comprising 104 patients with acute abdominal pain and 101 patients with dyspepsia, were drawn from the files of the Centre Hospitalier Universitaire in Sherbrooke. The history and physical-examination sheet was analyzed, coded and tested against the Leeds data base on a WANG 2200 computer, and the results were compared with the final Sherbrooke pathologic diagnosis. Overall the computer made a correct diagnosis in 78.8% of cases of acute abdominal pain and 70% of cases of dyspepsia. Computer diagnoses of appendicitis were correct in 97% of cases and the system recognized 91% of the actual appendicitis cases. Similar figures for cholecystitis were 91% and for peptic ulcer, 87%. However, the "pick-up" rate by the computer of pancreatitis was only 25%. It is concluded that geographical differences in disease presentation will probably not impair the validity of the computer method used in this study. A comparison of various diagnostic methods and levels of competence will await a prospective trial of this method.

Experience with computer-aided diagnosis of "dyspepsia" in a consecutive prospective series of 212 patients coming to surgery is described. Analysis is concentrated upon 122 patients who presented to an outpatient clinic de novo for diagnosis. During their first (outpatient) hospital contact, a firm diagnosis was made in just over half of these patients (though where made, it was usually correct). After full investigation, the diagnostic accuracy (prior to operation) was 92.6%. Using data elicited solely from the house surgeon's interview at the time of admission, the computer's overall diagnostic accuracy was 87.7%. The cost of each new computer diagnosis was around 25 new pence ($0.60). and the time taken was about 5 minutes. In a further small series designed to discriminate between organic and functional dyspepsia, the computer correctly assigned all but 1 of 23 patients with organic disease to the correct disease category. However, almost half of 33 patients with x-ray negative dyspepsia were predicted by the computer to have organic lesions. Time alone will tell whether the computer is a better early predictor of eventual organic disease than currently available radiologic methods.

By 1986 the central analysis team of this on-going multinational survey had received a total of 10,682 cases for analysis and had accepted 10,320. In all, some 26 centres in 17 countries, involving over 200 doctors, had participated in this survey. A common protocol was used for data collection; around 98% of all possible data was recorded (using precirculated definitions) and analysed via a computer-aided system in Leeds, England. The construction and format of a series of computer-aided decision-support and teaching programs has been described in an earlier (1982) report. These programs are currently available/in use in 10 countries. The present report concentrates upon an update of current material collated for the survey, some demographic trends, and special subreports (as with IBD survey) concerning acute abdominal pain in children and elderly patients, together with some preliminary data on the value of leucocyte count in patients with suspected appendicitis.

This presentation reviews the U.K. experience of computer-aided diagnosis of acute abdominal pain--which now relates to over 30,000 cases seen in more than 10 hospitals during a 13 years period. Following a discussion of the philosophy, construction and mode of usage of the systems employed, results of this experience are presented. Computer-aided diagnosis in this area has been shown to be feasible and (if correctly utilised) leads to improvements in patient care, diagnosis, and decision making by the doctors involved. In this context, the computer is simply one element of an integrated package reaffirming the importance of traditional clinical medicine.

This preliminary communication describes the initial results of a further special study investigating the disease spectrum and clinical presentation in a total of 1080 children admitted to hospital with acute abdominal pain (677 from the Children's Hospital, Sheffield, England, and the remaining 403 from hospitals in Paris, Oslo, Copenhagen, and Deventer). The disease spectrum in children differs radically from that in adults, well over 90% of cases being due to either acute appendicitis or non-specific abdominal pain (NSAP). The clinical presentation of both appendicitis and NSAP was found to differ from that in older age groups. These findings imply clearly that the use of the existing OMGE database for computer-aided diagnosis--using data drawn from cases of all ages--may not be optimal in children. A fresh database (using only children's data) was therefore created and tested. Its sensitivity for appendicitis was equivalent to that of inexperienced clinicians (79.6% versus 77.3%). The computer's specificity (over 70%) was higher than that of clinicians (52.7%). The findings also re-emphasise the value of disciplined data collection, and the implications for teaching are discussed.

This paper reviews the problems and prospects involved in providing computer-aided decision support in clinical medicine. First, the evaluation of medical innovation is discussed. It is suggested that there are three criteria by which an innovation may be judged, namely (1) a need for the innovation, (2) the ability of the innovation to fulfil that need and (3) the ability to do so without transgressing practical, ethical or legal boundaries. These problems are addressed in turn. The paper suggests, taking one area of clinical medicine as an example (acute abdominal pain) there is a clear need for decision support--since the area is not handled well by doctors in current practice. Evidence is adduced to suggest that the computer can provide decision support and do so without transgressing professional, ethical or legal boundaries. The obstacles to progress, which stand in the way of widespread implementation are briefly discussed. These are lack of medical terminology, poor man-machine interface and above all a lack of co-ordination. Finally, it is suggested that the most valuable facet of current systems is the discipline and precision in data collection they impose upon practicing doctors.

326 patients presenting with acute abdominal pain to a hospital in Tampere were compared with others in England and Norway, and with a large series of 6097 cases collected under the auspices of the World Organization of Gastro-Enterology. The distribution of disease in these 326 Finnish patients bore remarkable similarities to the distribution in other countries. When subjected to a computer-aided diagnostic analysis, comparing Finnish patients with UK and world-wide data, as regards appendicitis, cholecystitis and nonspecific pain, the computer performed roughly as well as the clinicians diagnosing the same cases. It is concluded that these diseases therefore have a common presentation in Finland with that elsewhere. As regards small bowel obstruction however, the computer diagnosed only 22% of cases correctly, (compared with the accuracy of clinical diagnosis 73%). This leads us to conclude that this disease presents unusually in Scandinavia, and possible reasons are discussed.

This paper describes the use of a system of computer aided diagnosis in an unselected, prospective survey of 393 women suffering from lower abdominal pain of less than 1 week's duration. An accurate diagnosis was made by clinicians at first patient contact in 68-5 per cent of the group of patients. The computer's diagnostic prediction (based on the same data) matched the final diagnosis in 81-6 per cent of the patients. During this survey a marked improvement in diagnostic accuracy was observed amongst the junior clinicians. It is suggested that this is because of the discipline of data collection imposed and the intermittent feedback received, and also that this educational aspect of computer usage may be of wider benefit.

This paper reports a comparison in Airedale District General Hospital between computer-aided diagnosis of dyspepsia and endoscopy in a prospective unselected series of 165 patients. Patients were interviewed immediately before endoscopy and the findings analysed by a small desk-top computer-aided system. Each 'new' patient was compared by the computer with a group of 360 similar patients from Leeds (25 miles away). Overall, 83% of the positive lesions found at endoscopy were correctly predicted by the computer, including all but three of the 22 cases of gastric cancer. It is suggested (1) that there is little loss of accuracy in transferring the computer-aided system from one locality to another; and (2) that a computer-aided analysis of the patient interview may be of value in selecting 'high-risk' patients for intensive investigation.

This paper presents a study of the diagnosis of "dyspepsia" in 154 patients based on data collected at their initial outpatient attendance via an interview with a non-medically qualified physician's assistant. The reactions of patients to this type of interview were favourable, and the data recorded were as reliable as those recorded by clinicians. We conclude (1) that the data recorded by the physician's assistant are valuable diagnostically; (2) where these cannot be collected by a qualified physician, this task may be delegated to a non-medically qualified person; but (3) this interview should augment and not replace the traditional clinical interview.

The OMGE multinational survey of patients with upper gastrointestinal bleeding demonstrated that it was possible to predict patient outcome, using a computer and a database of information from many centres. It remained to be seen whether this database could be used in specific remote areas of the world. To answer this question, two areas have been studied, Sikkim and China. In Sikkim, when the computer-aided prognostic system was used, 69% of patients put into a high-risk group for rebleeding actually did so; and 54% died. By contrast, only 2% of patients placed into a low-risk group for rebleeding did so. As there is little computer technology in Sikkim, a simplified scoring system was developed which gave the same predictive accuracy as the computer system. In China there was a slightly lower accuracy with both systems. Hence a new database and scoring system were created, using only information from Chinese patients. This database improved the results. From the studies it is suggested that these types of systems can be of value to patients in remote areas by targeting patients at high risk rebleeding or dying, so that the scarce resources available can be best used.

A multicentre study of computer aided diagnosis for patients with acute abdominal pain was performed in eight centres with over 250 participating doctors and 16,737 patients. Performance in diagnosis and decision making was compared over two periods: a test period (when a small computer system was provided to aid diagnosis) and a baseline period (before the system was installed). The two periods were well matched for type of case and rate of accrual. The system proved reliable and was used in 75.1% of possible cases. User reaction was broadly favourable. During the test period improvements were noted in diagnosis, decision making, and patient outcome. Initial diagnostic accuracy rose from 45.6% to 65.3%. The negative laparotomy rate fell by almost half, as did the perforation rate among patients with appendicitis (from 23.7% to 11.5%). The bad management error rate fell from 0.9% to 0.2%, and the observed mortality fell by 22.0%. The savings made were estimated as amounting to 278 laparotomies and 8,516 bed nights during the trial period--equivalent throughout the National Health Service to annual savings in resources worth over 20m pounds and direct cost savings of over 5m pounds. Computer aided diagnosis is a useful system for improving diagnosis and encouraging better clinical practice.

Physicians take both diagnosis and prognosis into account when allocating treatment. However, by "prognosis" physicians usually imply a somewhat vague impression concerning large groups of patients. One possible task for decision support studies is to design and construct systems that accurately predict individual patient prognoses. The authors constructed and tested such systems in three areas of medicine (inflammatory bowel disease, upper gastrointestinal tract hemorrhage, and acute chest pain). In each area, the individual patient's symptoms were compared with a computer-held database of information via a Bayesian analysis, prior and conditional probabilities being derived from large-scale real-life surveys. Prospective trials designed to test these predictive systems by reference to test series comprising over 4,000 patients indicate that a firm prognostic prediction can generally be made; where made, the accuracy of prediction is over 90%. Ways in which this type of prediction may be of clinical value are discussed.

This paper reports a survey of human and computer-aided diagnosis in a prospective consecutive series of 301 patients admitted to the hospital with lower gastrointestinal tract disease. At initial outpatient contact (at which time endoscopy was customarily performed), the clinicians' diagnostic accuracy was 64.5%. After biopsy, radiology, and other investigative procedures, the clinicians' preoperative diagnostic accuracy rose t0 82.7%. When data from the house surgeon's case notes were fed into the computer, its diagnostic prediction proved accurate in 77% of the 301 patients, and when details of the outpatient endoscopy were added to the house surgeon's case notes, the computer's diagnostic accuracy rose to 84.7%. It is suggested that computer-aided analysis may have a limited part to play in discriminating between the common causes of lower gastrointestinal disorders in routine clinical practice.

This paper reports the findings in a study involving 1537 patients with 'acute abdominal pain' presenting over a 13-month period to the Accident and Emergency Department of the General Infirmary at Leeds. Of these, 341 patients who proved to have pain of greater duration than a week, pain incidental to some other identifiable condition or no pain at the time of their attendance were excluded. The remaining 1196 were diagnosed clinically (using a structured case sheet) and subsequently by a Bayesian computer system. Feedback of the results of clinical and computer systems was given to clinicians at regular intervals. Clinical diagnostic accuracy in patients with surgical disorders rose from 40 per cent before the study to 61 per cent. Computer accuracy in these patients was 69-9 per cent. The proportion of patients sent home without ill effects rose from 20 per cent to 39 per cent. In other areas (e.g. gynaecology) the effects were less marked. It is suggested that the introduction of a simple postgraduate educational service, aided by a small computer, might prove of practical benefit in this clinical situation.

A simplified version of a system for computer-aided diagnosis of acute abdominal pain has been tested by "new" personnel unfamiliar with the previous system. After a two-month learning period the system proved more accurate in its diagnoses than the unaided clinician, and during the first five months of using the system the unaided clinicians' accuracy rose from 73% to 84%. When computer "feedback" was withdrawn the clinicians' diagnostic accuracy reverted towards the previous, "unaided" level. These findings further validate the concept of the computer as a potentially valuable diagnostic aid but indicate that a training period and computer feedback are important factors in its 4 use.

To assess the additional prognostic value of endoscopic stigmata of recent haemorrhage (SRH) in addition to clinical data, a pilot study was conducted on 207 duodenal ulcer patients from the OMGE survey. The incidence of SRH was compared in 145 patients who settled and 62 who re-bled. Only active bleeding emerged as significantly commoner in the latter group. Two computer-assisted predictions of further bleeding (before and after addition of the stigmata to a clinical database) were compared with the final outcome. Little improvement, in terms of prognosis for further bleeding, was obtained when SRH data were added. Similar findings were observed in a smaller series from Marburg, FRG.

This paper reports a controlled trial of human and computer-aided diagnosis in a series of 552 patients with acute abdominal pain. The overall diagnostic accuracy of the computer-aided system was 91.5% and that of the senior clinician to see each case was 81.2%. However, the clinician's diagnostic performance improved markedly during the period of the trial. The proportion of appendices which perforated before operation fell from 36% to 4% during the trial, and the negative laparotomy rate dropped sharply. After the trial closed in August 1972 these figures reverted towards their pretrial levels.It is suggested that while computer-aided diagnosis is a valuable direct adjunct to the clinician dealing with the "acute abdomen," he may also benefit in the short-term from the constant feedback he receives and from the disciplines and constraints involved in communicating with the computer.

This paper reports a comparison between two modes of computer-aided diagnosis in a real-time prospective trial involving 472 patients with acute abdominal pain. In the first mode the computer-aided system analysed each of the 472 patients by referring to data previously collated from a large series of 600 real-life patients. In the second mode the system used as a basis for its analysis "estimates" of probability provided by a group of six clinicians. The accuracy and reliability of both modes were compared with the performance of unaided clinicians.Using "real-life" data the computer system was significantly more effective than the unaided clinician. By contrast, when using the clinicians' own estimates the computer-aided system was often less effective than the unaided clinician-especially when diagnosing less common disorders. It seems, firstly, that future systems for computer-aided diagnosis should employ data from real-life and not clinicians' estimates, and, secondly, that clinicians themselves cannot analyse cases in a probabilistic fashion, since often they have little idea of what the "true" probabilities are.