**FORECASTING** (Greek. *prognosis* anticipation) — development of the forecast of a condition of the studied object according to information which is saved up by the present moment. The item makes a basis of any purposeful activity of the person; adoption of any decision of people carries out actually on the basis of P.

Mnozhestvo of tasks in medicine it is possible to group in several directions. One of them is P. of development of scientific research (see. Research institutes, planning and coordination of a scientific medical research in the USSR ). P. of indicators of health care belongs to other direction. E.g., experience of fight against flu epidemics allows to predict incidence for every period of next year and on this basis to take necessary measures, such as immunization of the population, accumulation of bank of necessary drugs, scheduling of a reshaping of bed fund, etc. Methods P. actively use in the organization of health care. E.g., scheduling of separate institutions allows to use more fully and rationally shots, bed fund, pharmaceuticals, a medical ambulance, placement of divisions, training of specialists, etc. (see. Planning of health care ).

In the course of medical providing troops (see. Meditsina military ) an important task is P. of number struck that is necessary during the calculation of volume of forces and means for assistance to victims, carrying out medical sorting and medical evacuation. The state struck is at the same time predicted that is important at establishment of priority of rendering the medical help to them.

Very large role is played by P. in a wedge, practice. On the basis of the knowledge, experience, qualification and data of inspection the doctor predicts a condition of the patient and defines tactics of treatment and observation (see. Forecast ). In the choice of a method of treatment of the patient also plays P.'s role as at purpose of the mode, diet, medicamentous therapy, physiotherapeutic procedures, etc. the doctor, considering, in addition to objective symptomatology, age of the patient, his profession, anamnestic data, existence of associated diseases, etc., chooses that complex to lay down. influences which shall lead to the most bystry positive take.

Problem solving of P. often demands the large volume of the calculations which are carried out by means of modern high-speed electronic computers (see. Electronic computer ).

Modern mathematical methods (see), the solutions of these tasks which are been the basis, it is possible to divide conditionally by types of the used information into three basic groups: methods of mathematical modeling (see), extrapolation and expert.

In a wedge, the practician P., based on mathematical modeling by means of the COMPUTER, allows, quantitatively estimating some symptoms and syndromes, to predict effect of treatment. E.g., the predictive analysis of changes of a hemodynamics in development and treatments of early stages of an idiopathic hypertensia allowed to establish what to lay down. the effect of adrenoblockers like propranolol is the best at hyperkinetic type of a hemodynamics and negative at hypokinetic. Methods of assessment of predictive influence of a pain syndrome, an idiopathic hypertensia and some other factors for the outcome of an acute myocardial infarction are developed.

In certain cases the forecast is under construction on the basis of a small amount of observations (so-called group of training). This forecast can serve as the recommendation to the doctor in his practical activities.

For use of methods of extrapolation it is necessary to have so-called temporary ranks, i.e. the numerical indicators describing the predicted phenomenon during observations (experiments). At creation of schedules of changes of these indicators extrapolation consists throughout curves for the predicted period (proceeding from preservation in this period of the developed tendencies).

A number of problems of P. is solved on the basis of mathematico-statistical methods of processing of expert information. For their use it is necessary to describe P.'s problem and initial information in the established form. The description begins with drawing up the list of entrance and output variables. Output variables are meant as those indicators which are required to be predicted, and under entrance — indicators, on the Crimea there are statistical data or data obtained as a result of objective inspection of the patient (temperature, pulse, etc.), and also the indicators chosen by the researcher or the doctor (e.g., the size of the category at a countershock, a type of medicine and its dosage, etc.). The first type of entrance variables contains information, on to-ruyu the researcher does not influence (these variables call uncontrollable). The second type of entrance variables contains information, to-ruyu the researcher (doctor) chooses; these variables call managed.

Thus, the description begins with allocation of entrance uncontrollable variables X and the entrance managed variables and. Forecasting consists in assessment of output variables U at preset values of X and and. If Z — quantitative assessment At, then P. consists in creation of dependence of Z from X and and, i.e. finding of function Z (X, U).

For difficult objects to determine precisely what will be At at set by X and and, usually happens it is impossible, i.e. discrepancy between Wu Yi his assessment by Z will always be. There is a problem of the choice of such forecast which is most exact and guarantees the smallest mistake. For establishment of dependence of Y from X and also use statistical and expert data. As dependence At from X and 17 has uncertain character, enter a concept of probability of this or that forecast. Use the device of mathematical statistics to determination of probability (see. Probabilities theory ).

The probability of existence infark that a myocardium’ depending on a combination of these or those symptoms is shown in table 1 made on the basis of the expert information obtained from 11 doctors (experts).

Apparently, 11 experts (doctors) give various data. Expert information generalizes the statistical data which are saved up by the doctor in the course of work. Therefore, naturally, that data of each new expert supplement the previous data. In the 1st column of 8 experts always in the practice in the presence of these symptoms at patients revealed a myocardial infarction, at the same time three experts had cases when the myocardial infarction did not appear. It means that the forecast of existence of a myocardial infarction cannot be considered absolute, i.e. equal 1,0. In this case as the elementary forecast it is necessary to choose a half-sum of the minimum and maximum probability, i.e. 1/2 (1,0+ 0,9) — 0,95. Accuracy of this forecast is equal + 0,05. The corresponding results and for other columns are given in the lower line of table 1. E.g., the patient has a combination of the symptoms given in the 7th column. We determine probability of a myocardial infarction by the table. It is equal 0,35, and the accuracy of the forecast 0,25, i.e. accuracy is comparable with the probability. Therefore value 0,35 is doubtful and it can be strongly underestimated. If to take the maximum probability, then it is equal to 0,6. Conclusion: for specification of the diagnosis additional information is necessary. In other case (see the 9th column) the accuracy of the forecast is comparable with the probability again, but the maximum value of probability of existence of a myocardial infarction is equal to 0,8. Therefore, uncertainty and danger of inexact diagnosis in this case are much higher, than in previous therefore it is necessary to begin treatment of the patient as at a myocardial infarction and at the same time urgently to obtain additional information. In case the patient has a symptom complex given in the 1st column, intensive observation and treatment need to be begun at once, already at a pre-hospital stage since the probability of existence of a myocardial infarction is very high (0,95), and the mistake makes only 0,05, i.e. is very small in comparison with probability.

In tables 2a and 26 expert estimates of probability of a favorable outcome of a myocardial infarction are given in the acute period depending on symptoms and the choice of pharmaceuticals. Apparently, even the same expert could not manage the indication of one probability. In the lower line of the table 2a indicated values of average probabilities of a favorable outcome, and also the accuracy of these values is specified. In the analysis of the table it becomes clear that, napr, according to the 7th column, the probability of a favorable outcome is high and therefore, at the specified symptomatology the choice of the given therapy is justified. At the same time, according to the 3rd column, the choice of pharmaceuticals cannot be considered effective and it is necessary to carry out more active therapy. According to the ninth and tenth columns, use of the given drugs is unreliable, however not only medicinal therapy, but also a lack of information on the patient can be the cause of a failure. Really, the accuracy of the forecast is comparable in these cases with the probability and for specification of the forecast it is necessary to conduct additional researches.

It should be noted that at the heart of mathematical methods P. on the basis of statistical and expert information different statistical approaches lie. When uncertainty of situations, with to-rymi faces the researcher, is too big, the role of the game methods supplementing and developing classical statistical methods increases. E.g., each expert (doctor) draws a conclusion on the basis of the experience and statistical data, and processing of expert information can be made by game methods as it was shown on the given examples.

Systems in which process of collection of information, its input in the COMPUTER, processings and deliveries of results is completely automated (see are created and are successfully operated. Identification ), what gives big help to the doctor, especially during the rendering the urgentny help to the patient.

See also Diagnosis machine , Forecasting epidemiological .

**Bibliography:** Breydo M. D., Neymark Yu. I. and Durnovo A. N. Minimax algorithms of training and classification, in book: Automation, the Organization, Diagnosis, under the editorship of V. V. Pa-rin, p.1, page 231, M., 1971; Computing systems and automatic diagnosis of heart diseases, the lane with English, under the editorship of Ts. Cáceres and L. Dreyfus, M., 1974; The Myocardial infarction, under the editorship of E. Kordeya and X. J. K. Svona, the lane with English, M., 1977; Kamensk E. I., Martynov 3. And. and Alekseeva 3. M. Experience of use operational kvali - a metric technique for medico-geographical assessment of a malyariogennost of territories, Medical parazitol., t. 47, No. 5, page 61, 1978; Machine diagnosis and information search in medicine, under the editorship of A. A. Vishnevsky, etc., M., 1969; R I would be L. B., Saakian V. G. and Yashin A. I. About one way of forecasting of indicators of a health system, Automatic equipment and telemechanics, No. 12, page 125, 1978; Simonov P. V., Anisimov S.A. and Raybman N. S. Game approach during the processing of physiological data on the example of a research of emotional reaction of the person, Zhurn. vyssh. nervn. deyateln., t. 28, century 4, page 675, 1978; D. K Falcons. Mathematical modeling in medicine, M., 1974; Halfen E. Sh. and Zaferman D. M. The forecast of outcomes of a myocardial infarction, in book: Use matemat. methods in izuch. cordial vessel. patol., under the editorship of E. Sh. Hal-fen, page 3, Saratov, 1971.

*S. A. Anisimov, V. N. Raybman.*