APPLICATION OF PHYSICIANS PREDICTIONS OF MEAL AND EXERCISE EFFECTS ONBLOOD-GLUCOSE CONTROL TO A COMPUTER-SIMULATION

Our aim was to develop a computer simulator program that allows patien ts to practise insulin dose and dietary adjustment on a day of planned exercise, and shows 4he resulting blood glucose response in an averag e diabetic patient. The degree of blood glucose change predicted by th e program was determined from changes predicted by five local speciali sts in seven hypothetical scenarios involving exercise +/- dietary or insulin dose adjustments. The program was then tested against 18 outsi de specialists' responses in 7 different scenarios. The program simula tes the 24 h glycaemic response after 45 min mild or moderate exercise starting 2 h after meals, as well as changes to this response induced by alterations in dietary carbohydrate and/or insulin dose. Coefficie nts of variation of specialists' blood glucose predictions were greate r for exercise (35% local, 31% outside specialists) than dietary chang e (7% local, 10% outside specialists; p = 0.002-0.04). The program's p redicted change in blood glucose levels in the seven scenarios correla ted well with the outside specialists' corresponding mean predictions (r = 0.97; p = 0.0001). We conclude that specialists are less consiste nt in predicting glycaemic change with exercise than with dietary alte ration. Nevertheless it is possible to represent their predictions in a computerized simulator for diabetic patient education.