[The Bayesian statistical theory in the diagnosis of malignant and non-malignant diseases of the lung, pleura and mediastinum].

The Bayesian algorithm was used to assess the probable diagnosis in 1262 patients with a recently diagnosed finding on X-rays of the chest and the results were compared with the final diagnosis. The patients were with regard to the X-ray picture divided into 9 groups: hilar, solitary, multiple, segmental, non-segmental, cavity, diffuse, pleural and mediastinal lesions. Using the Bayesian algorithm, commonly accessible factors were processed: age, sex, case-history, cigarette smoking, red cell sedimentation rate, number of leucocytes and diameter of solitary parenchymatous lesions and the impact of these factors for assessment of probability of a malignant or non-malignant lesion was evaluated. The reliability in different X-ray lesions was within the range of 84.2% to 92.4%. The authors evaluated also tests of sensitivity, specificity, the reliability of forecast of a positive and negative result, which confirmed the differences in the different groups which showed evaluated. Analysis of the results, provided evidence that the Bayesian algorithm is a promising objective method for the forecast of a malignant or non-malignant diagnosis in patients with a newly diagnosed X-ray finding of the chest.