Computer Program for Diagnosing and Teaching Geographic Me Dic Ine Stephen A

Computer Program for Diagnosing and Teaching Geographic Me dic ine Stephen A. Berger and Uri Blackman

One of the unique aspects of infectious disease is its neighboring countries and previous years when neces- wide variety, both in time and place.The specialist prac- sary (Table 1). In cases where the accuracy of disease ticing in India may have little or no expertise in Peru- reporting was suspect (e.g.,AIDS in Africa), more real- vian disease.A colleague in NewYork may be called upon istic published estimates were used. to diagnose and treat conditions originating in Africa,Asia, The data base is limited to infectious diseases South America, Fiji and Papua, New Guinea. At the (Table 2). It does not include slow viral illnesses and a same time, this colleague must be familiar with the number of self-defined and obvious conditions such as pathogens that originate in Texas, Hawaii, and Canada. otitis externa and furunculosis.As the program is designed Indeed, even the full-time infectious diseases specialist may to diagnose clinically apparent disease, data regarding not be conversant in diseases such as lagochilascariasis, asymptomatic carriage or infestation were adjusted louping ill, and lobomycosis. War, famine, education, accordingly. Figures regarding the incidence of signs and immigration, and business travel have contributed to the symptoms within each specific disease were derived advent of specialists in Geographic Medicine and Empo- from standard textbooks and reviews. Clinical and epi- riatrics, otherwise known as Travel medicine. demiologic data are updated on a continous basis. The “art” of diagnosis is largely an ability (albeit sub- The program user is first requested to indicate the conscious) to rank probabilities based on the incidences country of disease origin and is then presented with a of likely diseases and the chance of encountering given list of 22 basic clinical parameters, which are grouped clinical features within each disease. In theory, Bayesian according to body system.A + or - response to each of analysis could be employed to diagnose disease accurately the latter is indicated by using any of a variety of com- when given proper input.A multicenter study was under- puter keystr0kes.A + response automatically opens a com- taken to test a comprehensive computer driven-soft- puter window that requests further details.Thus, if the ware program that incorporates worldwide epidemiologic user indicates that a rash is present, he will be asked to and clinical parameters. further define the nature and distribution of the skin lesions. An additional window is available for the entry Materials and Methods of laboratory test results (hematologic, cerebrospin, hepatic, or renal) if available. Computer Program Design User input is processed by a Bayesian matrix, and Interactive data bases that represent rates and clin- compatible diagnoses are presented in order of probability ical probabilities were constructed for 308 diseases; 127 in a bar graph and numerical format.Ancillary clues for symptoms, signs, and laboratory findings; and 205 coun- all listed diseases are accessed by specified key strokes as tries. Reported statistics published by the World Health follows: incubation period, clinical hints, geographic dis- Organization and national health ministries were used tribution, vector, vehicle, reservoir, etc.Additionally,drugs where available.These were supplemented by data for of choice and dosages for adult or pediatric therapy are 1isted.The diagnosis list is accompanied by an ancillary screen, which indicates rare (albeit compatible) clinical findings in each disease listed for the patient in question. Stephen A. Berger, MD, The Infectious Diseases Division, An additional interactive screen lists all additional clin- Tel-Aviv Sourasky Medical Center, and Uri Blackman, BS, The Department of Computer Science, University of Tel-Aviv, ical findings that could improve diagnostic specificity. Tel-Aviv, Israel Separate computer modules allow the user to study Participating study institutions: Soroka Medical Center, Beer specific diseases and antiinfective agents without regard Sheva [Pediatrics and Internal Medicine]; Edith Wolfson to a specific patient.The user may, for example, request Medical Center, Holon; Haemek Hospital, Afula; Chaim a listing for all parasitic diseases acquired in Togo from Sheba Medical Center, Tel Hashomer; Carmel Hospital, Haifa the bites of mosquitoes; or of all drugs which interact with Reprint requests: Stephen A. Berger, MD, Dept. of alcohol. In addition to the epidemiologic and clinical para- Microbiology, Tel Aviv Medical Center, 6 Weitzman Street, meters outlined above, screens are available that outline Tel Aviv 64239, Israel the worldwide distribution of each disease, as well as the J Travel Med 1995; 2199-203 current status ofAIDS, malaria, tuberculosis,yellow fever,

199 200 Journal of Travel Medicine, Volume 2, Number 3

Table 1 Sources Used in Maintaining The Epidemiologic Database Official Health Ministry Reports Archives of Internal Medicine Bericht Uber dat Gesundheitswesen in Osterreich British Melcal Journal [Austria] Bulletin of the World Health Organization Boletin Epidemiologico de Chile Clinical Infectious Diseases Boletin Epidemiologico Nacional [Argentina] Clinical Microbiology Reviews Boletin Epidemiologico y Microbiologico [Spain] European journal of Microbiology and Infectious Boletin lnformativo [Bolivia] Diseases Bulletin Epidemiologique Hebdomadaire [France] Harefuah Canada Communicable Disease Report [Canada] Infectious Disease Clinics CDR Weekly [United Kingdom] Infectious Disease Clinics of North America Choroby Zakazne I Zatrucia W Polsce [Poland] International Journal of Systematic Bacteriology Communicable Diseases Intelligence [Australia] Israel Journal of Medical Sciences Community Health & Disease Surveillance News Letter JAMA [Oman] journal of Antimicrobial Chemotherapy Comportamiento de Patologias Immunoprevenibles journal of Clinical Microbiology [Argentina] Journal of Clinical Pathology Daten des Gesundheitswesens [Germany] Journal of Hospital Infection EpidAktuellt [Sweden] Journal of Internal Medicine Epidemiology Bulletin [Taiwan] Journal of Infectious Diseases EPI-NYT [Denmark] Journal of Pediatrics Heilbrigdisskyrslur [Iceland] Journal ofTravel Medicine IASR Uapan] Lancet Health Statistics Ireland [Ireland] Medical Journal of Australia Monthly Epidemiological Bulletin [Israel] Medicine Morbidity and Mortality Weekly Report [USA] Morbidity and Mortality Weekly Report (CDC) MSIS-rapport [Norway] New England Journal of Medicine Notiziario dell’Instituto Superiore de Sanita [Italy] Pediatric Clinics of North America Terveys [Finland] The Pediatric Infectious Disease Journal Weekly Epidemiological Record [WHO] Pediatrics Journals and Periodicals Reviews of Infectious Diseases AIDS Scandinavian Journal of Infectious Diseases American Journal of Clinical Pathology South African Medical Journal American Journal of Diseases of Children Southern Medical Journal American Journal of Epidemiology The Medical Letter American Journal of Medicine Transactions of the Royal Society ofTropical Medicine American Journal of Public Health and Hygiene American Journal ofTropical Medicine and Hygiene Tubercle Annals of Internal Medicine World Health Statistics Quarterly Antimicrobial Agents and Chemotherapy Applied Microbiology

and cholera. The therapeutic spectrum, toxicity, dosage results were collated and entered into a data base (dBase and other characteristics of anti-infective agents and III+) prior to to review of the clinical diagnoses. vaccines are also available. Statistical analysis employed the chi-square test for unpaired proportions. Multicenter Study Questionnaires reflecting the computer input screen Results were distributed to six senior full-time infectious disease specialists. (The authors’ own institution was Four hundred ninety and five of 513 cases submit- excluded). Participants were requested to record all pos- ted were suitable for analysis (Table 3). Ninety four itive and negative clinical data for consecutive patients individual infectious diseases were represented among with established diagnoses. Since the majority of cases these cases (Table 2).The computer program accurately were anticipated to represent disease acquired in the study identified the clinical diagnosis in 75.3% of actual cases country (Israel) a similar number 0f“hypothetical” cases and in 64.0% of hypothetical cases (p = .009).The clin- acquired abroad was also elicited. Questionnaires were ical diagnosis was included in the computer differential assigned code numbers and submitted in a blinded fash- diagnosis list in 94.7%. The accuracy of diagnosis was ion, with diagnoses recorded on a separate sheet.AU highest for parasitic disease (p = .04) and diseases acquired Berger and Blackman, Computer Program for Diagnosing & Teaching Geographic Medicine 201

Table 2 Diseases and Pathogens Included in the Data Base Abscess, intraabdominal* Cutaneous larva migrans" Herpes simplex infection* Mycobacteriosis - M. Actinomycosis Cutaneous leishmaniasis* Herpes simplex encephalitis* ulcerans Adenovirus infection Cyclospora infection Herpesvirus simiae infection Mycobacteriosis - systemic* Aeromonas & marine Cysticercosis* Herpes zoster* Mycoplasma pneunioniae Vibrio infx. Cytomegalovirus infection* Heterophyiasis infec.* Myiasis* AIDS* Dengue* Histoplasmosis* Nanophyetiasis Amebiasis* Derniatophytosis Histoplasmosis - African Necrotizing skidsoft tissue Amoeba - free living* Dicrocoeliasis HIV infection - initial ink. Angiostrongyliasis IXentamoebal diarrhea illness* Nocardiosis* Angiostrongyliasis abdominal Ilioctophyme renale Hookworm North Asian tick typhus Anisakiasis infection Hymenolepis diminuta Nonvalk agent gastroenteritis Anthrax* Diphtheria infection O'nyong nyong Argentine hemorrhagic fever Diphy llobothr iasis Hymenolepis nana infection Ockelbo disease Ascariasis Dipylidiasis Ilheus Oesophagostomiasis Aseptic meningitis, viral* Dirofilariasis Influenza* Omsk hemorrhagic fever Aspergillosis Dracunculiasis* Intracranial venous Onchocerciasis* Babesiosis* Eastern equine encephalitis thrombosis Opisthorchiasis Bacillary angiomatosis Ebola Disease Isosporiasis Orbital infection Bacillus cereus food Echinococcosis* Japanese encephalitis* Orf poisoning Echinococcus - multilocular Karelian fever Ornithosis" Balantidiasis Echinococcus vogeli Kawasaki disease Oropouche Bartonellosis infection Kingella infection Osteomyelitis Bertielliasis Echinostomiasis Kyasanur Forest disease Otitis media Blastocystis honiinis infection Ehrlichiosis - E. Lagochilascariasis Paracoccidioidomycosis Blastomycosis* chaffeenensis* Laryngotracheobronchitis Paragonimiasis Bolivian hemorrhagic fever Ehrlichiosis - E. sennetsu Lassa fever* Parainfluenza virus infection Botulism Endemic syphilis (bejel) Legionellosis* Parvovirus B19 infection Brain abscess* Endocarditis - infectious* Leprosy Pasteurellosis* Brazilian purpuric fever Entamoeba polecki infection Leptospirosis* Pediculosis Brucellosis* Enteritis necroticans Linguatulosis Penicilliosis California encephalitis group Enterobiasis Listeriosis* Pentastomiasis Campylobacteriosis* Enterovirus infection* Liver abscess, bacterial* Pericarditis, bacterial Candidiasis Entomophthoromycosis Lobomycosis Perinephric abscess* Capillariasis, hepatic Epidural abscess Loiasis* Perirectal abscess* Capillariasis, intestinal Erysipelas or cellulitis Louping ill Peritonitis, bacterial Cat scratch disease* Erysipeloid Lyme disease* Pertussis Cercarial dermatitis Erythrasnia Lymphocy tic Pharyngeal 81 cervical Chancroid* Escherichia coli diarrhea choriomeningitis space infx. Chikungunya European tick encephalitis Lymphogranuloma venereum Pharyngitis, acute bacterial Chlamydia infections, misc. Far eastern tick-borne Malaria* Pinta Chlamydia pneumoniae enceph. Malignant otitis externa Plague* infection Fascioliasis Mansonelliasis - M. ozzardi Plesiomonas enteritis Cholecystitis 81 cholangitis* Fasciolopsiasis Mansonelliasis - M. perstans Pleurodynia Cholera* Filariasis - Brugia malayi Mansonelliasis - M. Pneumocystis pneumonia* Chromomycosis Filariasis - Brugia timori streptocerca Pneumonia, bacterial* Chronic fatigue syndrome Filariasis - Bancroftian Marburg virus disease Pogosta disease Chronic meningococcemia Gardnerella vaginalis Mayaro Poliomyelitis (wild or Clonorchiasis* infection Measles* vaccine) Clostridial food poisoning Gastrodiscoidiasis Mediterranean spotted fever* Powassan Clostridial myonecrosis Giardiasis* Melioidosis* Pseudocowpox Clostridium difficile colitis* Glanders Meningitis, bacterial* Pyodermas (impetigo, Coccidioidomycosis Gnathostomiasis* Metagonimiasis abscess, etc) Coenurosis Gonorrhea Microsporidosis Pyomyositis* Colorado tick fever Granuloma inguinale Monkeypox Q fever* Common cold* Group C viral fevers Mononucleosis, infectious* Queensland tick typhus Conjunctivitis inclusion Hantavirus resp. distress* Mucocutaneous leishmaniasis Rabies Conjunctivitis viral Hemorrhagic fever & Mumps Rat bite fever - spirillary cowpox renal synd. Murray Valley encephalitis Rat bite fever - Crimean Congo Hepatitis A* Mycetoma streptobacillary hemorrhagic fever Hepatitis B* Mycobacteriosis - M. Relapsing fever* Cryptococcosis* Hepatitis C marinurn* Respiratory syncytial Cryptosporidiosis* Hepatitis delta infection Mycobacteriosis - M. infection Cyclospora infection Hepatitis E scrofulaceum Reve's svndrome 202 Journal of Travel Medicine, Volume 2, Number 3

Table 2 Diseases and Pathogens Included in the Data Base [Continued] Rheumatic fever Schistosomiasis - iaoonicum Tetanus Typhus - epidemic _.& Rhinoscleroma Schistosomiasis - mansoni* T helaziasis Typhus - scrub Rhinosporidiosis Schistosomiasis - mattheei Thogoto Toxic shock Urinary tract infection* Rhodococcus equi infection Schistosomiasis - mekongi syndrome* Varicella* Rickettsialpox Septicemia, bacterial* Toxocariasis* Venezuelan equine RiftValley fever Septicemia, fungal* Toxoplasmosis* encephalitis Rocio Shigellosis* Trachoma Venezuelan hemorrhagic Rocky Mountain spotted Sindbis Trench fever fever fever* Sinusitis Trichinosis* Vesicular stomatitis disease Roseola or human Smallpox Trichonioniasis Vibrio parahaeniolyticus herpesvirus 6 Sparganosis Trichostrongyliasis infection Ross River disease Spondweni Trichuriasis Visceral leishnianiasis* Rotavirus infection* Sporotrichosis Tropical phagedenic ulcer Wesselbron Rubella* St. Louis encephalitis Tropical pulmonary West Nile Fever Sabia Staphylococcal food eosinophilia Western equine encephalitis Salmonellosis poisoning Trypanosomiasis - African Whipple's disease Sandfly fever* Strongyloidiasis* Trypanosomiasis - Wound infection Sarcocystosis Subdural empyema American* Yaws Scabies Suppurative parotitis Tuberculosis* Yellow fever Scarlet fever Syngamiasis Tularemia* Yersiniosis" Schistosomiasis - Syphilis* Tungiasis Zygoniy cosis* haematobiurn* Taeniasis Typhoid and enteric fever* Schistosomiasis - intercalatum Tanapox virus disease Typhus - endemic *denotes cases submitted for diagnosis in the present study

in Africa (p = .04) and South East Asia (p = .03);no such diseases within any given country. Furthermore, the differences were noted with respect to body system and computer program assumes that the patient is a citizen patient age group. or local resident of the country in question. Incidence data for tourists and expatriates may vary from those of Discussion the indigenous population. In some cases, the country of acquisition may not match the country of residence. The major problem in developing an infectious dis- Selection of discriminative clinical and laboratory ease diagnosis program is difficulty in obtaining reliable parameters for the data base is complicated by the fact and accurate incidence data.The reporting rate for dis- that individual infections are quite sirmlar, often producing eases varies widely between countries and among differing fever, cough, rash, elevated white blood cell count, etc.

Table 3 Evaluation of a Computer-Driven Infectious Disease Diagnosis Program (percent) Nature oflnfection Bacterial Parasitic Viral Fungal Total Correct * Actual cases 150 30 100 15 295 222 (75.3) Hypothetical cases 97 60 33 10 200 128 (64.0) Total 247 90 133 25 495 Correct diagnosis* 186 (75.3) 60 (66.7) 88 (66.2) 16 (64.0) 350 (70.7) Correct diagnosis included in hfferentialt 236 (95.5) 87 (96.7) 124 (93.2) 22 (88.0) 469 (94.7)

Country of Acquisition

~~ ~ ~ ~~~~~ Israel Afnca Southeast Europe Latin North Other Asia Amenca America Number of cases 308 66 65 24 7 19 6 Correct diagnosis* 205 (66.6) 54 (81.8) 54 (83.1) 16 (66.7) 5 (71.4) 12 (63.2) 4 (66.7) Correct diagnosis included in differential+ 295 (95.8) 62 (93.9) 61 (93.8) 23 (95.8) 6 (85.7) 18 (94.7) 4 (66.7)

~~~ ~~ ~~~~~~~~~~~~~~ *Concordance between the correct clinical diagnosis and the disease listed first in the computer-generated differential diagnosis list; +the correct clinical diagnosis is included in the computer-generated diagnosis list Berger and Blackman, Computer Program for Diagnosing & Teaching Geographic Medicine 203

Similar abnormalities are also found in a variety of the expense of low “specificity” (ability to exclude irrel- noninfectious diseases. An additional difficulty in any evant diagnose^).^ Nevertheless, an accompanying edi- diagnostic program is the reliability of user input. The torial suggested that the alternative diagnoses listed were accuracy of clinical input is only as good as the accuracy often valuable to the clinician and would not otherwise of history taking, physical examination, and laboratory have been c~nsidered.~Indeed, when dealing with infec- testing. In some instances, more than one disease may be tion acquired in an exotic country, the clinician might present, or clinical observations may be factitious or find an exhaustive differential diagnosis to be quite help- unrelated to the present illness. In the current study, ful.Although only 94 diseases (30.5% of the program data actual cases were correctly diagnosed more often than base) were represented, our preliminary study suggests that hypothetical cases (e.g., acquired overseas), thereby sug- the program under study is comprehensive and accurate, gesting relative unfarmliarity of infectious diseases experts and could prove useful in the diagnosis of infectious with the clinical features of “exotic” diseases. and tropical disease. An expanded study among infectious During the period January 1989-February 1992, diseases physicians in the United States will be under- Index Medicus listed 2063 papers under the subject head- taken in the near future. ing, “Diagnosis, Computer Assisted,” and 7139 under the heading, “Software”; however, no program specifi- References cally designed for diagnosis in infectious and geographic medicine has been reported in the English language lit- 1. Aizenstein HJ. Computer systems for medical diagnosis. erature to date. Existing computer-driven diagnostic JAMA 1992; 267:166-170. programs have failed to adequately simulate human intel- 2. Waxman HS, Worley WE. Computer-assisted adult medical ligence or find widespread practical use in the field.’--? diagnosis: subject review and evaluation of a new As such, software systems (including the program under microcomputer-based system. Medicine (Baltimore) 1990; 69: 125-1 36. study) are marketed for “decision support,” and display 3. Szolovits P, Patil RS, Schwartz WS. Artificial intelligence in appropriate disclaimers that remind the user that clini- medical diagnosis. Ann Intern Med 1988; 108: 80-87. cal judgment still takes precedence over computer “expert 4. Uerner ES,Webster GO, Shugerman AA, et al. Performance systems.” of four computer-based diagnostic systems. N Engl J Med In a recent study, the “sensitivity” (i.e., ability to 1994; 330:1792-1796. include the correct diagnosis in a differential list) of 5. Kassirer JP. A report card on computer-assisted diagnosis - foursuch systems was found to be adequate, but often at the grade: C. N Engl J Med 1994; 330:1824-1825

Sundial, 30 meters high, at the observatory of Jantra Mantra built in 1728 by Jai Singh in Jaipur/Rajastan/lndia. Submitted by Danielle Gyurech, MD and Julian Schilling, MD.