Malocclusion
Conditions
Interventions
Source Domain Dataset (Study One):None
Target domain dataset (Study One):None
Paired dataset (Study One):None
Questioning Volunteers (Study Two):Raise orthodontic consultation questions
Scoring Expert Group (Study Two):Score the quality of the responses generated by LLM
Sponsors
Beijing Stomatological Hospital , Capital Medical University
Eligibility
Sex/Gender
All
Age
18 Years to 55 Years
Inclusion criteria
Inclusion criteria: 1. Source domain dataset (retrospective) : (1) The photo was taken of the permanent dentition, with no gender restrictions; (2) The Fusion One system has a complete anterior orthodontic treatment image. 2. Target domain and paired dataset (prospective) : (1) Age 18-55 years old, gender not limited; (2) Be capable of understanding the research procedures and cooperating to complete the collection of facial images (including mobile phone shooting or standard shooting simultaneously). 3. Volunteer Questions: (1) Age: 18-55 years old, gender not limited; (2) No background in stomatology; (3) Normal language expression and comprehension abilities. 4. Scoring Expert: (1) Possessing a senior professional title in orthodontics (associate chief physician or chief physician); (2) Have more than 5 years of clinical experience in orthodontics.
Exclusion criteria
Exclusion criteria: 1. Source domain dataset (retrospective) : None; 2. Target domain and paired dataset (prospective) : (1) Combined with severe systemic diseases (such as active malignant tumors, etc.); (2) There is severe oral infection, tumor or trauma that affects image interpretation; (3) Pregnant or lactating women; (4) Those who are undergoing orthodontic treatment; (5) People with mental or intellectual disabilities. 3. Questioning volunteers: (1) Pregnant or lactating women; (2) People with mental or intellectual disabilities. 4. Scoring experts: (1) Those who have conflicts of interest with the project team of this research; (2) Those who are unable to independently complete the scoring task within the prescribed time.
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| Accuracy; | — |
Secondary
| Measure | Time frame |
|---|---|
| AUC;Precision;F1 score;Recall;Dataset construction metrics: Sample size of each diagnostic label, image quality grading distribution, and completion rate of paired data collection.;Annotation quality indicator: Consistency among annotators (Cohen's Kappa coefficient).;The performance of large language models: The 3C score of the generated responses by experts (correctness, clarity, conciseness), and the consistency among raters (Spearman correlation coefficient); | — |
Countries
China
Contacts
Public ContactYang Kai
Beijing Stomatological Hospital , Capital Medical University
Outcome results
None listed