Fig. 7. The percentage of correct responses (true positives) and false positives for the main classes of sounds, depending on the assessment of their own auscultation skills for children (a, c) and adults (b, d). To emphasize the differences, the points for individual groups are connected by lines.
When analyzing the classes of crackles and wheezes, one can notice that the lowest scores were obtained by those who rated their skills the lowest but also those who rated their skills the highest. However, there was no statistically significant relationship between the assessment of their own skills and the level of correct or false detections of the auscultation phenomena (for all correlations p> 0.05).
In the test, both the reference sounds included on the CD of the physician handbook (Wilkins et al., 2004) and real-life sounds recorded using the Littmann 3200 stethoscope were used. The effect of the type of sound recording on the results was analyzed. The results for correct, partially correct, and incorrect detections were compared, taking into account the specialization of the participants. Based on the results of the Wilcoxon test, no significant difference in the percentage of any type of diagnoses between the two kinds of recording (CD and Littmann 3200) was observed in any of the participants (p> 0.05). This means that the type of recording did not affect the results obtained by the participants, and they perform similarly with the classification of sounds in the case of different origins.
In the next step, we analyzed what phenomena other than the standard ones were marked by subjects. It should be emphasized that the purpose of this analysis was first of all to draw attention to the grouping of classes, which may be the first step in further unification of the nomenclature of sounds. It was assumed that phenomena that are most similar would also most often be confused with each other. Therefore, the obtained relationships were grouped by classes and the analysis of those are depicted, namely: crackle class (Fig. 8), breath sounds (vesicular and louder) and bronchial sounds (Fig. 9), wheezes and rhonci (Fig. 10). This approach is in line with the assumptions of Pasterkamp et al. (2016) for other languages.
In the case of the juxtaposition of coarse, medium, fine crackles and crepitus (Fig. 8), it can be seen that these are classes that are confused. Coarse crackles are most often confused with other types of crackles and crepitus, and even rhonchi. In the medium crackle class the results are similar to those for coarse crackles, but the most noticeable additional class marked for this phenomenon was crepitus.