2/21 JDR

Question 1

Describe the study on machine learning and 3rd molar assessment

Carvalho et al JDR

Answer 1

Used AI to detect and classify 3rd molar proximity to the IAN nerve in panoramic radiographs.

4,516 panos used to train the machine learning device.
Classifications:
*3rd molar root development: Undeveloped, fully developed, or uncertain development
* IAN proximity: No superimposition, superimposition <50%, or superimposition >50%

120 panos were then processed through AI.

3 different AI models were used for 3rd molar detection:
* YOLOv5
* SDU-Net = had best performance (97.7% accuracy)
* U-Net

2 different AI models were used for IAN superimposition and root development:
* ResNet-101 = best performance (~94% for both)
* Vit-B

Question 2

Describe the study on how smoking exposure affects nonsurgical periodontal therapy - 1 year followup

Leite et al JDR

Answer 2

80 patients
* Exam, then SRP followed by perio maintenances every 3 months
* “Most” patients were Stage III Grade B

Compared light smokers / quitters, moderate smokers, and heavy smokers. (Couldn’t find in the article their definition for these terms)

RESULTS:
*Heavy smokers (overall?) had 10 more sites with ≥8mm CAL loss, compared with the light smokers
* The heavy smokers had only 50% reduction of # of sites, compared with the light smokers.
* Non-surgical therapy had no effect on heavy smoker severe periodontitis patients, compared with the light and moderate smokers.

NOTE: Somewhat poor study design and lack of clarity

Question 3

Describe the study on identifying dental implant systems using a large-scale multicenter data set.

Park et al JDR

Answer 3

Used the ResNet-50 AI system to identify dental implants in xrays.

Data set:
* 37,442 Periapical xrays and 113,291 panoramics.
* There were 25 different dental implant systems.

120,585 were used for training, 15,074 for validation, and the AI was tested on 15,074 xrays.

RESULTS:
The ResNet-50 AI was much more accurate than dental professionals at identifying implants:
* AI: 82% accurate
* Dental professionals: 24%
AI was also much faster and saved more than 81 minutes of time. (To read 200 xrays: AI: 4.5 minutes; Dental professionals: 91.3 minutes)

Question 4

Describe how TAM receptors are therapeutic targets in bone regeneration.

Decker et al JDR

Answer 4

Mouse study involving extraction and daily treatment with RXDX-106 (a TAM inhibitor). Histology at 7 days

Results:
* TAM inhibition doesn’t change immune dynamics after extraction.
* TAM inhibition accelerated alveolar bone healing by promoting osteogenic differentiation of alveolar bone mesenchymal stem cells. = 4X increase in activity.

Question 5

Describe the study on the efficacy of mouthwash containing CHX and CPC in Covid-19 positive patients.

Bonn et al JDR

Answer 5

61 COVID-positive patients total. Oropharyngeal samples were collected before, and 30 minutes after the rinse.
* Test group: rinse with CHX and CPC
* Control group: Placebo rinse

RESULTS:
* Viral load decreased in both groups after the mouthwash.
* Viral infectivity significantly decreased in the test group.
* SSD in viral infectivity was found in patients who had 3 vaccine shots, compared with 2 shots.

Question 6

Describe the study on clinical and radiographic outcomes of short (6mm) versus long (10mm) implants after 10 years of loading.

Sahrmann et al JDR

Answer 6

Radiographic study

• Short (6mm) implant survival: 86%
• Long (10mm) implant survival: 97%
• Survival was NSSD
• NSSD Marginal bone loss