Validation of ReadSmart4U: An Automated Tool for Neuropsychological Assessment Reporting.

Background: Neuropsychological assessments are critical to cognitive care, but are time-consuming and often of variable quality. Automated tools, such as ReadSmart4U, improve report quality and consistency while meeting the growing demand for cognitive assessments.

Methods: This retrospective cross-sectional study analysed 150 neuropsychological assessments stratified by cognitive diagnosis (normal cognition, mild cognitive impairment and Alzheimer's disease) from the Clinical Data Warehouse of a university-affiliated referral hospital (2010-2020). Reports were generated for each assessment by ReadSmart4U and certified clinical psychologists (CCPs). Three blinded CCPs assessed report quality using the Integrated Scoring for Quality Assessment of Structure (ISQAS) for absolute quality and the Integrated Scoring for Quality Comparison of Superiority (ISQCS) for paired comparisons. Domains assessed included terminology accuracy, interpretation accuracy, usefulness and writing quality.

Results: ReadSmart4U-generated reports outperformed CCP reports in all ISQAS domains, with mean overall quality scores of 87.3 ± 3.4 vs. 74.5 ± 6.7 (P < 0.001). Domain-specific scores were higher for terminology accuracy (31.4 ± 1.5 vs. 26.8 ± 2.1), interpretation accuracy (32.2 ± 1.7 vs. 27.3 ± 2.4), usefulness (10.8 ± 0.9 vs. 9.2 ± 1.2) and writing quality (14.5 ± 0.6 vs. 12.7 ± 1.1; all P < 0.001). In the ISQCS evaluations, ReadSmart4U reports were judged superior in 80.2% of cases for overall quality and 58.0-88.2% for domains (P < 0.001).

Conclusion: ReadSmart4U significantly improves the quality and consistency of neuropsychological assessment reports compared to CCPs, reducing workload and supporting the integration of automated tools into clinical workflows to improve cognitive care.