CAMEOS ALGORITHM
Introduction
Computer Assisted Medical Evaluation of Symptoms (CAMEOS) is an expert system programmed as a web application. The system has been conceptualized, designed and developed by the principal author. He has been a Board certified family physician and a consultant to Postgraduate Institute of Medicine University of Colombo Sri Lanka, He had been a member of the Board of Study in Family Medicine and an educator, examiner in postgraduate and doctoral level examinations in Family Medicine. His clinical experience in clinical primary care medicine is well over 30 yrs. The system has been in use in his primary care practice. It has undergone multiple iterations of debugging and development. Now it is in a fairly stable level of development. It is now used by 10-15 family physicians in private practices and in a Family Medicine unit attached to University of Jaffna. Currently several validation and usability studies are underway and system development is halted until further results are available from research in progress.
Description of the system :
A computer application to help primary care physicians (PCPs) diagnose common diseases seen in primary care. In addition it can also help PCPs to evaluate symptoms, symptom complexes, red flags and many other referral indications.
List the system functionality :
This is a web based application to be used by PCPs in their daily clinical work. It is to be used for live data entry by the user physician. The main program user interface consists of 3 main areas : for identification data entry, symptom panel and illness panel and a submit button. After the basic identification data is entered physician interviews the patient. Having interviewed the patient physician can enter the data. The symptoms are categorized into the body systems mainly. So the respiratory symptoms are under respiratory system and bowel symptoms are under GIT symptoms etc.
Physician clicks and select the symptoms from the panel. Then the items in the illness panel are selected. These comprise detailed evaluation of the patient's illness experience. This include among many other aspects, illness duration, course of the illness, severity, impact of the disease etc. Once this is done the physician user will submit the data. The data is saved and the program generates the output in another html file. This output is the statement of the diagnosis arrived by the system and the diagnostic reasoning behind the conclusion. Typically this runs to 7-10 pages of A4 paper size. The output itself is separated into many sections where the title headings include : Copyright and warranty issues, Symptoms reported, Illness data reported, Calculating the diagnostic statistics, Differential diagnosis, Rank ordering of diagnosis, Selection of the highest probability diagnosis by the System. The output can help the user physician to change the probabilities of the final diagnosis or the individual diseases of the list of differential diagnosis based on the information provided in
various sections of the output statement. The feedback of changing the input and the resulting
new probabilities are almost instantaneous.
Clinical Informatics Problems it is designed to address :
The problems of unacceptable levels of diagnostic errors in primary care, unacceptable levels of disease specific under-diagnosis and missed diagnosis, poor documentation of the clinical data, unusable free text data entry, lack of documentation of clinical reasoning and lack of documentation of differential diagnosis are several problems this system is designed to solve. In addition the well known inter-practice variations in diagnostic and management performances are also problems this type of tools can solve. The output file generated by the program which is a html file contains all the reasoning behind the programs clinical decision making. As the data is saved in a database it is available for later retrieval. In addition the html file which is generated can be converted into a pdf and given to the patient if required.
Innovative features of the proposed EMR app
Web based expert system, data entry during the clinical consultation, based fully on a hypothetico deductive type of reasoning, knowledge retrieval from the knowledge based based on a pattern matching algorithm mimicking physicians' natural diagnostic reasoning, documentation of entire diagnostic process, including the historical data collection, ability to change the diagnostic statistics with the change of data input, saving of data for later retrieval are some of the key features when compared with other ES currently available.