Artificial Intelligence in Orthodontic Smart Application for Treatment Coaching and Its Impact on Clinical Performance of Patients Monitored with AI-TeleHealth System

Abstract

Background: Treatment of malocclusion with clear removable appliances like Invisalign^® or Spark™, require considerable higher level of patient compliance when compared to conventional fixed braces. The clinical outcomes and treatment efficiency strongly depend on the patient's discipline. Smart treatment coaching applications, like strojCHECK^® are efficient for improving patient compliance.

Purpose: To evaluate the impact of computerized personalized decision algorithms responding to observed and anticipated patient behavior implemented as an update of an existing clinical orthodontic application (app).

Materials and methods: Variables such as (1) patient app interaction, (2) patient app discipline and (3) clinical aligner tracking evaluated by artificial intelligence system (AI) system-Dental monitoring^® were observed on the set of 86 patients. Two 60-day periods were evaluated; before and after the app was updated with decision tree processes.

Results: All variables showed significant improvement after the update except for the manifestation of clinical non-tracking in men, evaluated by artificial intelligence from video scans.

Conclusions: Implementation of application update including computerized decision processes can significantly enhance clinical performance of existing health care applications and improve patients' compliance. Using the algorithm with decision tree architecture could create a baseline for further machine learning optimization.

Keywords: AI; behavior change techniques; clear aligners; computerized learning; decision tree algorithm; orthodontic treatment; smart application; telemedicine.

Figure 1

Introduction of Dental Monitoring^® (DM): (a) Patient holding scan-box looking into the mirror, instructed by nurse, is scanning her first intraoral scan with her own mobile. (b) DM has its own app that is used for requests, uploading and reporting of the scans. Its first use is usually also instructed by nurse. The photo was taken for purposes of this article and is published with full written consent of the person.

Figure 2

Main screens of orthodontic smart-app StrojCHECK^®: (a) Main dashboard of the app provides complex view of the current day, remaining time, planned and executed activities, lower chart visualizes daily performance of time of aligners out-of-mouth, above it is current balance of earned points/treatment discount and current number of aligner (b) Remainders are frequently send as a push notifications to patient mobile and wearable either inquiring if the patient has ended an activity and forgot to return the aligners or is still performing the activity. (c) An activity can be finished by active patient interaction (with bonus) or automatically (with sanction) (d) Side menu allows user to personalize the app, set up scheduled routines, language, learn about his own performance or contact his doctor directly reporting an event.

Figure 3

Global schematics of technological platforms fundamental to the complex system of StrojCHECK^® from the back-end server through different dental clinics with their separate administrator portals up to end-user devices.

Figure 4

The view of the screen for administration. The portal of the server back-end for doctors and other administrators provides useful interface for statistical data processing and interactions with the system. Pie charts from left describe percentage of active users within the last 72 h (from all registered users), types of monitored appliances in the active users (Spark lite, Spark full, Invisalign teen, Invisalign lite, Invisalign comprehensive, Invisalign first), types of monitored appliances in the non-active users, types of daily limits of removed aligners (over 120 min, within 120 min limit and under 15 min). On the Figure bellow there are visualized user statistics, average times for various habits, Average time per aligner or frequency of patient reports.

Figure 5

Scheme of three related parts with decision processes that were implemented by the update and represent three layers of system-user interaction.

Figure 6

Screens of StrojCHECK^® app regarding incentives: (a) Patient view of gained points that equal total discount from the treatment budget (b) Achieved incentives(badges) and other badges to be achieved are introduced grayed, dynamically introduced by the system. (c) List of gained points by following the rules and fulfilling motivational bonuses offered by the system (as special motivational events).

Figure 7

Prediction interval and linear regression line for Interaction (after-before) vs. Age in the investigated sample (correlation coefficient (r) = 0.160; p = 0.1416). Each circle represents a data point. The red line is the regression line (the linear model) and the black lines show the 95% prediction band for the forecasted Interaction (after-before).

Figure 8

Prediction interval and linear regression line for Discipline (after-before) vs. Age in the investigated sample (r = 0.210; p = 0.0528). Regression line and 95% prediction interval are denoted as in Figure 7.

Figure 9

Prediction interval and linear regression line for No-go scans (after-before) vs. Age in the investigated sample (r = 0.0507; p = 0.6433). Regression line and 95% prediction interval are denoted as in Figure 7.

Figure 10

Prediction interval and linear regression line for Age vs. GO scans change (After-Before) vs. Age in the investigated sample (r = 0.0507; p = 0.6433). Regression line and 95% prediction interval are denoted as in Figure 7. Coefficient (r) = 0.089831 p = 0.4108.

Figure 11

(a) Differences before and after update regarding app interaction for all participants. (Paired t test of variable Discipline) Measured parameter represented patient app interaction in easy and mostly fun interactions including sharing on social networks or achieving interesting badges. (b) Differences before and after update regarding app interaction for all participants (Paired t test of variable Discipline). Measured parameter represented patient app interaction in difficult and regular way fulfilling required rules of disciplined use that included teeth and appliance cleaning twice a day (separated by 180 min), at least once per day eating and drinking and as well as fitting the aligner out-of-mouth time between 15 and 120 min. This parameter improved significantly as well.

Figure 12

Differences before and after the app update regarding patient clinical performance observed with dental monitoring for all participants. (Paired t test) (a) Dental monitoring evaluated GO scans focused on proper aligner tracking, here is not a significant difference. (b) Dental monitoring evaluated NO-GO scans focused on proper aligner tracking, here is a significant improvement.

Figure 13

Differences before and after the app update in NO-GO Dental monitoring scans in (a) Females (significant) (b) Males. Difference, despite were less frequent, is not significant.

Figure 14

Differences before and after the app update in GO Dental monitoring scans were not significant (a) in females (insignificant) (b) Males (insignificant).