Clinical and peristaltic-pump bench validation datasets for the AI-UFD study

This restricted dataset supports the study “Artificial Intelligence-Based Uroflowmetry Device: Prospective Clinical and Experimental Validation”. The record contains two analysis datasets used to reproduce the main statistical results of the study: (1) a de-identified participant-level clinical validation dataset and (2) a run-level peristaltic-pump bench validation dataset.

The clinical dataset contains de-identified measurements from 50 male participants with lower urinary tract symptoms who completed same-day standard gravity-based uroflowmetry device (SG-UFD) and artificial intelligence-based uroflowmetry device (AI-UFD) assessments. The dataset includes SG-UFD, AI-UFD, OnePlus, and Redmi measurements for key uroflowmetry parameters, including Qmax, Qave, voided volume, flow time, void time, and time to maximum flow, together with derived variables used for agreement, diagnostic-accuracy, and sensitivity analyses. Direct identifiers, raw patient videos, exact clinical encounter dates, and variables that could reasonably enable participant re-identification are not included.

The peristaltic-pump bench dataset contains run-level measurements from 60 bench experiments across five target flow levels and four waveform families. The dataset includes physical-truth measurements and corresponding AI-UFD, OnePlus, and Redmi outputs for Qmax and voided volume, supporting the bench agreement and condition-specific accuracy analyses.

These datasets are provided for research transparency, reproducibility assessment, and qualified methodological research. Because the clinical dataset derives from human participant data and because the AI-UFD platform is undergoing clinical-trial and institutional intellectual-property procedures, all files in this record are provided under restricted access.

Access request conditions:

Access to the files is restricted. Requests will be considered from qualified researchers for non-commercial research, methodological validation, and reproducibility assessment (please reach out to [e-mail]).  Requests must include the requester’s full name, institutional affiliation, institutional email address, research purpose, planned analyses, and confirmation that the data will not be used for participant re-identification, commercial product development, redistribution, or attempts to reconstruct proprietary AI-UFD software or model components.

Access may be granted at the discretion of the corresponding authors and is subject to institutional approval, applicable ethical and data-governance requirements, confidentiality obligations, and completion of a data-use agreement where required. The files may not be redistributed to third parties without written permission from the corresponding authors.