Trial Review

The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been endorsed by the ANZCTR. Before participating in a study, talk to your health care provider and refer to this information for consumers
Trial registered on ANZCTR


Registration number
ACTRN12621001020875p
Ethics application status
Submitted, not yet approved
Date submitted
22/04/2021
Date registered
4/08/2021
Date last updated
4/08/2021
Date data sharing statement initially provided
4/08/2021
Type of registration
Prospectively registered

Titles & IDs
Public title
Determining whether Deep Learning Analysis of Facial Imaging is Effective in Predicting Difficult Intubation
Scientific title
Predicting Anatomically Difficult Intubation Through Deep Learning Analysis of 3-Dimensional Facial Imaging of Patients in a Pre-Anaesthetic Assessment Clinic
Secondary ID [1] 304036 0
None
Universal Trial Number (UTN)
Trial acronym
Linked study record

Health condition
Health condition(s) or problem(s) studied:
Difficult intubation 321668 0
Airway management 322763 0
Condition category
Condition code
Anaesthesiology 319415 319415 0 0
Anaesthetics

Intervention/exposure
Study type
Observational
Patient registry
False
Target follow-up duration
Target follow-up type
Description of intervention(s) / exposure
Patients will be recruited from the perioperative anaesthetic assessment clinic at Royal Perth Hospital (RPH) over a 6-month period. Following patients' informed consent we will take a 3-Dimensional digital photographs of their face front on and side on. We will record basic demographics including age, gender, weight, and height. When the patient undergoes surgery the responsible anaesthetist will complete a data collection that assess the difficulty of their intubation. The timing of the 3D photographs in relationship to the surgery will be variable given the heterogeneous group of patients that are seen at the pre-anaesthetic clinic, but in general surgery is expected to follow around 1 to 3 months following image acquisition.

We will use this information to develop and to train a deep learning algorithm which uses patient demographics and 3D photograph as an input, and predict difficulty of intubation as an output.
Intervention code [1] 320357 0
Early Detection / Screening
Comparator / control treatment
Anaesthetic trainees and consultants clinical assessment of the predicted difficulty of intubation (reference comparator).
Control group
Active

Outcomes
Primary outcome [1] 327274 0
Deep learning model accuracy in classifying patients level of intubation difficulty. Accuracy will be assessed by comparing the number of difficult intubations identified by the deep learning model to number of difficulty intubations identified by the anaesthetist (prior to actual intubation).
Timepoint [1] 327274 0
Photographs to be used as input for deep learning model will be determined at baseline.
Surgery up to 6 months after end of patient recruitment
Secondary outcome [1] 394436 0
Nil
Timepoint [1] 394436 0
Nil

Eligibility
Key inclusion criteria
Adult patients undergoing elective surgery that are anticipated to require intubation.
Minimum age
18 Years
Maximum age
No limit
Sex
Both males and females
Can healthy volunteers participate?
No
Key exclusion criteria
Patients will be excluded if after enrollment they do not undergo intubation at surgery, their surgery is cancelled, or their data collection form is not completed by the treating anaesthetist.

Study design
Purpose
Screening
Duration
Longitudinal
Selection
Convenience sample
Timing
Prospective
Statistical methods / analysis
We will provide descriptive statistics on the characteristic of the dataset used. We will report the predictive performance of our models and anaethetists in terms of sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve. Confidence intervals and power calculations will be included where appropriate.

Recruitment
Recruitment status
Not yet recruiting
Date of first participant enrolment
Anticipated
1/09/2021
Actual
Date of last participant enrolment
Anticipated
1/01/2022
Actual
Date of last data collection
Anticipated
1/08/2022
Actual
Sample size
Target
250
Accrual to date
Final
Recruitment in Australia
Recruitment state(s)
WA
Recruitment hospital [1] 19176 0
Royal Perth Hospital - Perth
Recruitment postcode(s) [1] 33748 0
6000 - Perth

Funding & Sponsors
Funding source category [1] 308418 0
University
Name [1] 308418 0
University of Western Australia
Country [1] 308418 0
Australia
Funding source category [2] 308419 0
Hospital
Name [2] 308419 0
Royal Perth Hospital
Country [2] 308419 0
Australia
Primary sponsor type
Individual
Name
Dr Jonathon Stewart
Address
The University of Western Australia, 35 Stirling Highway, Perth WA 6009, Australia
Country
Australia
Secondary sponsor category [1] 309253 0
Individual
Name [1] 309253 0
Professor Thomas Ledowski
Address [1] 309253 0
The University of Western Australia, 35 Stirling Highway, Perth WA 6009, Australia
Country [1] 309253 0
Australia

Ethics approval
Ethics application status
Submitted, not yet approved
Ethics committee name [1] 308380 0
Royal Perth Hospital Human Research Ethics Committee
Ethics committee address [1] 308380 0
East Metropolitan Health Service Executive
Level 2, Kirkman House
198 Wellington Street
Perth Western Australia 6000
Ethics committee country [1] 308380 0
Australia
Date submitted for ethics approval [1] 308380 0
14/05/2021
Approval date [1] 308380 0
Ethics approval number [1] 308380 0

Summary
Brief summary
When a patient has surgery under general anaesthesia, it is one of the jobs of the anaesthetist to secure and maintain a patent airway. Placing a breathing tube into a patients’ windpipe is one of the most commonly performed procedures to secure the patient’s airway. Though this so-called intubation is usually an easy task for the anaesthetist, sometimes it can be difficult. This leads to a potentially life-threatening situation.

Hence, it is part of the routine preoperative anaesthetic assessment to examine a patients’ airway in order to attempt to predict how difficult it will be to intubate them. There are a number of examination techniques and tools that have been developed for this, but none are sensitive enough to rely on.

Deep learning algorithms are able to learn to map complex and subtle relationships between input variables to a known output. This relationship is learned from the data, and the algorithm can then be used to predict outputs for future inputs. Deep learning algorithms have been successfully applied to a wide range of computer vision tasks
This research will apply deep learning to patients basic demographics and photographs of patients face and neck in order to predict how difficult they will be to intubate. We will recruit patients from the perioperative anaesthetic assessment clinic at Royal Perth Hospital who are expected to require intubation for their surgery. We will take 3-dimensional stereophotographs of the patients’ face and neck in various positions. The difficulty of intubation will be recorded at the time of surgery. We train the deep learning model using simple patient data such as age, gender height, weight, and the images as an input, and the intubation difficulty as an output. We will then attempt to predict how difficult intubation will be for patients given their images as an input. We will also compare the results of the deep learning algorithm to anaethetists predictions.
Trial website
Trial related presentations / publications
Public notes

Contacts
Principal investigator
Name 110510 0
Dr Jonathon Stewart
Address 110510 0
The University of Western Australia, 35 Stirling Highway, Perth WA 6009, Australia
Country 110510 0
Australia
Phone 110510 0
+61 435211352
Fax 110510 0
Email 110510 0
[email protected]
Contact person for public queries
Name 110511 0
Dr Jonathon Stewart
Address 110511 0
The University of Western Australia, 35 Stirling Highway, Perth WA 6009, Australia
Country 110511 0
Australia
Phone 110511 0
+61 435211352
Fax 110511 0
Email 110511 0
[email protected]
Contact person for scientific queries
Name 110512 0
Dr Jonathon Stewart
Address 110512 0
The University of Western Australia, 35 Stirling Highway, Perth WA 6009, Australia
Country 110512 0
Australia
Phone 110512 0
+61 435211352
Fax 110512 0
Email 110512 0
[email protected]

Data sharing statement
Will individual participant data (IPD) for this trial be available (including data dictionaries)?
No
No/undecided IPD sharing reason/comment


What supporting documents are/will be available?

No Supporting Document Provided



Results publications and other study-related documents

Documents added manually
No documents have been uploaded by study researchers.

Documents added automatically
No additional documents have been identified.