Introduction:
Auditory brainstem response (ABR) is a key objective diagnostic method in audiology; however, the interpretation of recordings – especially estimating hearing thresholds – is challenging for novices. The aim of this preliminary descriptive study was to document the learning process of novices in ABR threshold interpretation in preparation for the ABR.AI project.

Material and methods:
Seven students with varying levels of baseline theoretical knowledge participated in the training. The program consisted of a multi-stage cycle including theoretical lectures, individual analysis of ABR recordings, independent practice with available expert consultation, and case-based discussions. The students’ hearing threshold estimations were compared with expert assessments across three clinical datasets (115–122 cases each). Results were presented as the percentage of full agreement (0 dB difference), minor discrepancies (≤ 10 dB), and large discrepancies (> 10 dB) relative to the expert’s assessment.

Results:
Across the three training stages, mean full agreement with the expert increased from 55% (Set 1) to 60% (Set 3), while major discrepancies decreased from 9% to 6%. Despite individual differences in learning dynamics, the majority of participants demonstrated a consolidation of interpretive skills and a reduction in large diagnostic errors. Most deviations were ≤ 10 dB, indicating that discrepancies were generally small and clinically acceptable.

Conclusions:
This preliminary descriptive study demonstrated that multi-stage, expert-guided training was associated with an improvement in ABR threshold estimation accuracy, and most importantly, with a clinically meaningful reduction in large discrepancies (>10 dB) from the expert’s assessment. These findings highlight the importance of structured, feedback-based learning in the education of audiology professionals.