Neovascular Age-Related Macular Degeneration (nAMD)
Conditions
Keywords
Decison-making, AI-Assisted treatment, artificial intelligence, clinical decision support, ophthalmology, SD-OCT, neovascular age-related macular degeneration, nAMD, anti-VEGF, explainable AI
Brief summary
This is a research plan from the University of Munich (LMU) that aims to study how the use of AI reports can impact ophthalmologists' decisions regarding treatment for patients with neovascular age-related macular degeneration (nAMD). This disease is a leading cause of vision loss, and while anti-VEGF treatments are effective, they require careful monitoring and retreatment decisions to maximize benefits. The study will involve up to 1000 ophthalmologists with varying levels of expertise. These ophthalmologists will review SD-OCT scans and make treatment decisions before and after reviewing AI-generated reports. The primary objective is to compare these decisions and see how the AI reports influence them. Secondary objectives include assessing the accuracy and safety of the AI reports.
Detailed description
This research project at LMU delves into the intersection of artificial augmentation and ophthalmology, specifically focusing on how AI-generated 2nd opinion reports can aid in the treatment planning of neovascular age-related macular degeneration (nAMD). The project will involve a diverse group of up to 1000 ophthalmologists, categorized into six user groups based on their expertise, ranging from residents to seasoned retina specialists. The core of the research involves assessing the impact of AI-generated 2nd opinion reports on ophthalmologists' treatment decisions for nAMD. Participants will review SD-OCT scans and make initial treatment decisions. Subsequently, they will review AI-generated reports for the same scans and have the opportunity to revise their decisions. This process aims to evaluate the influence of AI insights on clinical judgment. The project will be conducted virtually, with participants enrolling online from various countries. Data collection will be facilitated through an electronic system, ensuring efficiency and security. Statistical analysis will primarily involve descriptive statistics to summarize the findings. The results of the study will be disseminated through publication in a peer-reviewed journal.
Interventions
BEHAVIORALAI assisted assessment of SD-OCT scans
AI 2nd opinion report on nAMD treatment planning
Sponsors
Deepeye Medical GmbH
Technomics Research
M3 Macula Monitor Muenster
Johannes Schiefelbein
Study design
Allocation
NA
Intervention model
SINGLE_GROUP
Primary purpose
BASIC_SCIENCE
Masking
NONE
Intervention model description
Survey
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
* Electronically consented to the informed consent form (eICF) * Criteria to be included in one of the following six Ophthalmology user groups: Group 1 Non-retina specialist Group: Ophthalmology, completed ophthalmology residence with no or another subspecialty other than retina (e.g., Glaucoma, refractive, etc) Group 2 Resident Group: \<5 years in residency in ophthalmology Group 3 Fellow Group: Retina specialist in training Group: in fellowship in vitreoretinal medicine, medical retina Group 4 Retina specialist Group: completed retina training, regular requalification Group 5 Junior reader Group: have already gained experience in the reporting clinical routine with the diagnostics in question and completed the initial certification process at an Image and Reading Center (acc. to centre's SOP) Group 6 Senior reader Group: specialist with several years of experience in the relevant field or have completed at least 3 years of residency training. Completed the certification process at the Image and Reading Center (acc. to centre's SOP)
Exclusion criteria
* Not an Ophthalmologist. * Does not have time to participate in the estimated project duration of 30 minutes.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Ophthalmologist's Treatment Decision | There is only one survey filled out by the participant. In this survey, only one time point when participant views the SD-OCT and AI 2nd opinion report and fills out the survey questions. | The number and percentage of initial treatment decisions that stayed the same after their review of the AI-CDS report The number and percentage of initial decisions that changed after their review of the AI-CDS report |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Performance Accuracy | There is only one survey filled out by the participant. In this survey, only one time point when participant views the SD-OCT and AI 2nd opinion report and fills out the survey questions. | The number and percentage of correct assessments (accuracy) made by AI-CDS report vs image grading/reading center (M³ Macula Monitor Münster) assessment (control / gold standard / ground truth) The percentage of correct assessments (accuracy) made by each of the 6 user group vs. image grading and reading center (M³ Macula Monitor Münster) assessment (control / gold standard / ground truth). How often is the AI-CDS report correct? How often is each of the 6 user groups correct after viewing just the SD-OCT image? Intra-rater reliability: How often are ophthalmologists (each user group) right/wrong after viewing AI-CDS report and made any decision changes? |
| Safety prediciton assessment | There is only one survey filled out by the participant. In this survey, only one time point when participant views the SD-OCT and AI 2nd opinion report and fills out the survey questions. | Yes/No prediction rates If an ophthalmologist is correct and AI-CDS report incorrect, how often does AI-CDS report mislead? Whether false-positive or false-negative? If an ophthalmologist is wrong and AI-CDS report correct, how often does AI-CDS report correct? Whether false-positive or false-negative? |
| Exploratory AI-CDS report Impact on Decision Making | There is only one survey filled out by the participant. In this survey, only one time point when participant views the SD-OCT and AI 2nd opinion report and fills out the survey questions. | impact of AI-CDS report on ophthalmologist's decision-making survey |
Countries
Germany
Contacts
Primary ContactClinical Project Manager
Outcome results
None listed