Periodic Reporting for period 1 - orthomobile (A System for the Early Diagnosis of Skeletal Orthodontic Malocclusions)
Okres sprawozdawczy: 2023-08-01 do 2024-07-31
The project aims to develop AI-based software for free preliminary screenings of children as young as five globally, targeting early identification of orthodontic deformities with high accuracy for timely and cost-effective intervention.
Early detection of skeletal orthodontic problems allows for more effective and economical treatment, often avoiding invasive surgeries. This is crucial for socioeconomically disadvantaged children, ensuring access to necessary orthodontic care. Early treatment reduces the psychological and social impacts, such as stigmatisation and discrimination, and addresses potential health issues like sleep apnea, which can lead to neurocognitive and psychosocial problems if untreated.
The project's AI-based software achieves an 83.3% accuracy rate in detecting Class III malocclusions, surpassing the diagnostic accuracy of orthodontic specialists. The American Association of Orthodontics recommends checking every child before age 7. Early detection is essential as the upper jaw stops growing at 12, while the lower jaw continues until 20. Orthodontists' traditional methods often diagnose too late, leading to costly double jaw surgeries. Early intervention can correct jaw positions at around €300 per patient, while double jaw surgery costs about €10,000. The global market for these surgeries is valued at $1.5 billion.
A small lower jaw (Class II skeletal malocclusion) can narrow the airway, leading to sleep apnea and potential misdiagnoses of ADHD. Early diagnosis and treatment can prevent these issues, improving academic performance and reducing stigmatization. Boys with Class II malocclusion face societal biases associating strong jawlines with masculinity, affecting their psychological well-being. Media portrayals of individuals with Class III malocclusions can reinforce negative stereotypes. Early detection and treatment mitigate these issues, particularly during sensitive developmental periods.
Early diagnosis reduces the treatment accessibility gap among children and alleviates the burden on healthcare systems in countries with healthcare coverage, promoting social equity by reducing disparities in access to healthcare.
The project has received national and international patents and plans to raise public awareness about the importance of early detection of skeletal malocclusions. Collaborations with organizations like UNICEF and paediatric associations will enhance community engagement and ensure widespread utilization of the screening tool.
Early detection of skeletal orthodontic problems allows for more effective and economical treatment, often avoiding invasive surgeries. This is crucial for socioeconomically disadvantaged children, ensuring access to necessary orthodontic care. Early treatment reduces the psychological and social impacts, such as stigmatisation and discrimination, and addresses potential health issues like sleep apnea, which can lead to neurocognitive and psychosocial problems if untreated.
The project's AI-based software achieves an 83.3% accuracy rate in detecting Class III malocclusions, surpassing the diagnostic accuracy of orthodontic specialists. The American Association of Orthodontics recommends checking every child before age 7. Early detection is essential as the upper jaw stops growing at 12, while the lower jaw continues until 20. Orthodontists' traditional methods often diagnose too late, leading to costly double jaw surgeries. Early intervention can correct jaw positions at around €300 per patient, while double jaw surgery costs about €10,000. The global market for these surgeries is valued at $1.5 billion.
A small lower jaw (Class II skeletal malocclusion) can narrow the airway, leading to sleep apnea and potential misdiagnoses of ADHD. Early diagnosis and treatment can prevent these issues, improving academic performance and reducing stigmatization. Boys with Class II malocclusion face societal biases associating strong jawlines with masculinity, affecting their psychological well-being. Media portrayals of individuals with Class III malocclusions can reinforce negative stereotypes. Early detection and treatment mitigate these issues, particularly during sensitive developmental periods.
Early diagnosis reduces the treatment accessibility gap among children and alleviates the burden on healthcare systems in countries with healthcare coverage, promoting social equity by reducing disparities in access to healthcare.
The project has received national and international patents and plans to raise public awareness about the importance of early detection of skeletal malocclusions. Collaborations with organizations like UNICEF and paediatric associations will enhance community engagement and ensure widespread utilization of the screening tool.
The whole software and business management has been transferred to Atlassian Jira Work Management software system. Jira is installed on all programmers' and orthodontist's mobile phones to efficiently track the tasks between the company leaders and the programmers.
Our manuscript is published at a SCI and open-access Q1 journal on Jan 2024 ( https://doi.org/10.1016/j.jds.2023.05.001 )
We retrained our machine learning models using a doubled input patient size of adult+kids photos on kids. We validated the new model on an independent dataset of 60 kids patients diagnosed by different independent orthodontic experts. The newly trained model increased the accuracy by 6% to 83.3% and reduced the false positive rate to 12%.
We migrated the Turkish orthodontists' addresses, phone information, and ML models from university servers to Amazon Cloud servers. We enhanced the security of the cloud data storage by creating distinct AWS IAM roles for all software developers, including the project leaders, and reset the root password. We ensured that no one, including the project owners, used the root password in daily operations to increase security.
We added the statistical confidence level of the predictions to the backend API output. This confidence level will be incorporated into the front-end report to the clients.
We also experimented with training the ML model for long-face patients using the Mediapipe library facial points.
Our manuscript is published at a SCI and open-access Q1 journal on Jan 2024 ( https://doi.org/10.1016/j.jds.2023.05.001 )
We retrained our machine learning models using a doubled input patient size of adult+kids photos on kids. We validated the new model on an independent dataset of 60 kids patients diagnosed by different independent orthodontic experts. The newly trained model increased the accuracy by 6% to 83.3% and reduced the false positive rate to 12%.
We migrated the Turkish orthodontists' addresses, phone information, and ML models from university servers to Amazon Cloud servers. We enhanced the security of the cloud data storage by creating distinct AWS IAM roles for all software developers, including the project leaders, and reset the root password. We ensured that no one, including the project owners, used the root password in daily operations to increase security.
We added the statistical confidence level of the predictions to the backend API output. This confidence level will be incorporated into the front-end report to the clients.
We also experimented with training the ML model for long-face patients using the Mediapipe library facial points.
An investigation was conducted to determine how early diagnosis can reduce treatment costs. The number of face masks sold in Turkey between 2023 and 2024 was 552, including sales to the state. These devices cost an average of €20, and the treatment fee per person is €300 in private offices and free in public institutions. In contrast, double jaw surgery costs around €10,000. Our practice can potentially reduce the long-term healthcare burden associated with late-diagnosed malocclusions.
Early intervention can mask or correct physical symptoms, reducing bullying and stigmatization in schools. This is especially beneficial for boys with small and girls with prominent lower jaws, who often face teasing and societal biases. The app can be expanded to include functions like treatment outcome prediction and personalised treatment plans. Our long face detection study for the early diagnosis of mouth breathing has begun. If successful, correct myofunctional exercises can be recommended early, avoiding future issues such as maxillary transverse deficiency, mandibular retrognathia, periodontal problems, and high caries incidence.
Based on customer insights from interviews conducted in the U.S. we plan to implement a successful Market Expansion and Commercialization Plan for greater updates and success:
Customers are willing to pay a co-pay of $25-$50 if insurance companies cover the rest.
Reluctance to pay more than $10 out-of-pocket for AI diagnosis unless covered by insurance.
Feedback from a real doctor on diagnosed malocclusions via email is valued at around $15.
Including the cost of our app’s services in insurance coverage will enhance expansion.
We plan to introduce a basic service where users can receive skeletal diagnosis feedback from a real doctor via email for $5 to $50, depending on the orthodontist chosen. This service will launch within the next six months. Compliance with FDA and HIPAA regulations will increase user trust.
Campaigns will highlight success stories and testimonials from early users, focusing on raising public awareness. We will continue to publish open-access articles in high-impact dental journals to showcase our scientific and technical superiority. While focusing on B2B2C, we need to continue nurturing organic growth on the B2C. By leveraging both B2B and B2C channels, we aim to reach a breakeven point by the third year.
Early intervention can mask or correct physical symptoms, reducing bullying and stigmatization in schools. This is especially beneficial for boys with small and girls with prominent lower jaws, who often face teasing and societal biases. The app can be expanded to include functions like treatment outcome prediction and personalised treatment plans. Our long face detection study for the early diagnosis of mouth breathing has begun. If successful, correct myofunctional exercises can be recommended early, avoiding future issues such as maxillary transverse deficiency, mandibular retrognathia, periodontal problems, and high caries incidence.
Based on customer insights from interviews conducted in the U.S. we plan to implement a successful Market Expansion and Commercialization Plan for greater updates and success:
Customers are willing to pay a co-pay of $25-$50 if insurance companies cover the rest.
Reluctance to pay more than $10 out-of-pocket for AI diagnosis unless covered by insurance.
Feedback from a real doctor on diagnosed malocclusions via email is valued at around $15.
Including the cost of our app’s services in insurance coverage will enhance expansion.
We plan to introduce a basic service where users can receive skeletal diagnosis feedback from a real doctor via email for $5 to $50, depending on the orthodontist chosen. This service will launch within the next six months. Compliance with FDA and HIPAA regulations will increase user trust.
Campaigns will highlight success stories and testimonials from early users, focusing on raising public awareness. We will continue to publish open-access articles in high-impact dental journals to showcase our scientific and technical superiority. While focusing on B2B2C, we need to continue nurturing organic growth on the B2C. By leveraging both B2B and B2C channels, we aim to reach a breakeven point by the third year.