Al-based versus human-led training for bronchoscopy
| ISRCTN | ISRCTN63799884 |
|---|---|
| DOI | https://doi.org/10.1186/ISRCTN63799884 |
| Sponsor | University of Milano-Bicocca |
| Funder | Università degli Studi di Milano-Bicocca |
- Submission date
- 30/01/2026
- Registration date
- 06/02/2026
- Last edited
- 05/02/2026
- Recruitment status
- No longer recruiting
- Overall study status
- Completed
- Condition category
- Surgery
Plain English summary of protocol
Background and study aims
Residents in anesthesia and intensive care of the University of Milano-Bicocca without prior exposure to specific bronchoscopy training were enrolled to compare the AI-based to classical human-led training. Participants were assessed using the modified Bronchoscopy Skill and Task Assessment Tool (BSTAT) to evaluate the theoretical knowledge regarding the recognition of proximal bronchial anatomy (28 points total) and a practical component, assessing procedural positioning, airway wall trauma, correct intrabronchial scope position, and access to several tracheobronchial structures (27 points total). Of note, the proportion between the score driven by knowledge and practical component is similar to the original version of the BSTAT. Finally, similarly to the original BSTAT, the time required to complete the examination was recorded.
Who can participate?
Adult residents in anesthesia and intensive care.
What does the study involve?
Consent for the publication of data was obtained from residents. After a 1-hour frontal lecture on bronchoscopy and bronchial anatomy, the baseline bronchoscopy skills of all participants were tested using the BSTAT. Participants were thereafter randomized in a 1:1 ratio using sealed envelopes. The first group received classical training performed by an expert bronchoscopy instructor. The second group performed unsupervised training using the AI-based image recognition software. Each resident had 20 minutes of individual training and watched the individual training sessions of the other residents of her/his group. At the end of the training, each resident repeated the modified BSTAT. The assessment of the modified BSTAT was always performed by the same person, blinded to group allocation. Both baseline and post-training BSTAT examinations were conducted individually to prevent any learning effect from observation, ensuring that each resident's performance was based solely on their own training experience.
What are the possible benefits and risks of participating?
Benefits and risks not provided at time of registration
Where is the study run from?
University of Milan-Bicocca (Università degli Studi di Milano-Bicocca), Italy.
When is the study starting and how long is it expected to run for?
February 2024 to March 2024
Who is funding the study?
University of Milan-Bicocca (Università degli Studi di Milano-Bicocca), Italy.
Who is the main contact?
Prof Thomas Langer, thomas.langer@unimib.it
Contact information
Principal investigator, Scientific, Public
University of Milan-Bicocca, Monza, Italy; Department of Anesthesia and Intensive Care Medicine, Niguarda Ca' Granda
Milano
20162
Italy
| Phone | +39 0264448580 |
|---|---|
| thomas.langer@unimib.it |
Study information
| Primary study design | Interventional | |
|---|---|---|
| Allocation | Randomized controlled trial | |
| Masking | Open (masking not used) | |
| Control | Active | |
| Assignment | Crossover | |
| Purpose | Educational- training | |
| Scientific title | Artificial intelligence-based image recognition in bronchoscopy: a randomized controlled trial for training evaluation in intensive care residents | |
| Study acronym | AI-BRITE Trial | |
| Study objectives | The primary aim of this study is to compare the performance of residents in flexible bronchoscopy after specific training, either AI-based or human-led. Specifically, we hypothesize that the Bronchoscopy Skill and Task Assessment Tool (BSTAT) scores of residents undergoing AI-based training are similar to those assigned to human-led training. | |
| Ethics approval(s) | Ethics approval not required | |
| Health condition(s) or problem(s) studied | Anesthesia Resident | |
| Intervention | Participants are randomized in a 1:1 ratio using sealed envelopes. The first group receives classical training conducted by an expert bronchoscopy instructor. The second group performs unsupervised training using AI-based image recognition software. Each resident has 20 minutes of individual training and watches the individual training sessions of the other residents in her/his group. At the end of the training, each resident repeats the modified BSTAT. | |
| Intervention type | Other | |
| Primary outcome measure(s) |
| |
| Key secondary outcome measure(s) |
| |
| Completion date | 01/03/2024 |
Eligibility
| Participant type(s) | |
|---|---|
| Age group | Mixed |
| Lower age limit | 18 Years |
| Upper age limit | 99 Years |
| Sex | All |
| Target sample size at registration | 22 |
| Total final enrolment | 22 |
| Key inclusion criteria | 1. Residents in anesthesia and intensive care 2. Without prior exposure to specific bronchoscopy training 3. Accepted to participate |
| Key exclusion criteria | Not meeting the key inclusion criteria |
| Date of first enrolment | 01/02/2024 |
| Date of final enrolment | 29/02/2024 |
Locations
Countries of recruitment
- Italy
Study participating centres
Results and Publications
| Individual participant data (IPD) Intention to share | No |
|---|---|
| IPD sharing plan summary | Not expected to be made available |
| IPD sharing plan |
Editorial Notes
05/02/2026: Study’s existence confirmed by the Head of the Department of Medicine and Surgery of the University of Milano-Bicocca, Italy.
