LLM-Powered Treatment Plans: A New Era in Dental Care
Q&M Dental Group’s LLM-Powered Treatment Plans: A New Era in Dental Care
As the Director of AI Innovation at AI Singapore, I’ve always believed in the transformative power of artificial intelligence (AI) across industries. Our latest collaboration with Q&M Dental Group, a leading dental healthcare provider in Asia, is proof of that. Together, we’ve developed an LLM-Powered AI module called “Treatment Plan,” built for the dental industry.
The Power of AI in Dentistry
The “Treatment Plan” module uses Large Language Models (LLMs) to analyze patient data, including X-rays and dental records, and to generate personalized treatment plans. It combines the strengths of conventional clinical decision support systems with the processing capabilities of LLMs, prioritizing precision, reliability, and patient-centered care so that each patient receives the most suitable treatment recommendations.
Benefits for Patients and Dental Professionals
Patients are the ultimate beneficiaries. The module enables early detection of oral health issues, provides a range of treatment options based on individual needs, and empowers patients to make informed decisions about their dental care. The iterative platform continuously learns and adapts, so patients receive the most up-to-date and relevant recommendations.
Dental professionals benefit too. The “Treatment Plan” module streamlines the diagnostic process, reducing the time spent analyzing X-rays and generating treatment plans. That lets them focus on delivering superior care, enhancing both clinical outcomes and the overall patient experience.
A complete patient journey
With the “Treatment Plan” module, Q&M Dental Group now offers a complete AI-powered digital dentistry experience, from initial consultation to X-ray analysis, pathology detection, treatment recommendations, and a personalized dental health report. Integrating AI across the journey improves efficiency and patient satisfaction.
A Collaborative Effort for the Future of Dental Care
The “Treatment Plan” was developed through AI Singapore’s Short Industry Project (SIP) program. EM2AI, a subsidiary of Q&M, brought its domain expertise in dental care; AI Singapore brought AI engineering resources and support.
This project shows AI Singapore fulfilling its mission to anchor deep national capabilities in AI, creating social and economic impact. Partnering with industry leaders like Q&M improves patient outcomes and transforms how dental professionals approach treatment planning.
Meet the AI Engineer Behind the “Treatment Plan”
The “Treatment Plan” module was spearheaded by Mr. Benjamin Phang, an AI engineer at AI Singapore. Benjamin did not start with a computing background. Only 20% of our apprentices hold a degree in computer science; the rest come from a variety of disciplines. With the right mindset and training, one can succeed in AI.
Here are Benjamin’s thoughts:
Can you tell me a bit about your background and how you became interested in AI?
I am a physicist by training and throughout my studies I was exposed to AI in different forms during my internships. For example, we implemented reinforcement learning to stabilize a laser in a lab, and during my internship at CERN, I came to know of the TrackML challenge. However, outside of that, I have been inspired by the OpenAI Five which was once again a Reinforcement Learning Project.
What motivated you to join AI Singapore’s AI Apprenticeship Program (AIAP)?
My first job post-graduation was that of a data engineer in a startup, however, as is expected in a startup, I wore many hats and one of those was a data analyst. However, as I was doing my data analyst job, I felt like there must be something better, something more current that I could provide. It was at that moment that a friend introduced me to AIAP, and I felt like my prayers were answered. A structured program which gave me skills that were vetted through the industry was exactly what I wanted, and I found it through AIAP.
How did your experience in the AIAP prepare you for the “Treatment Plan” project?
Through the deep-skilling phase in the AIAP, we learnt the basics about natural language processing (NLP), from cleaning input text by removing stop words to the various tokenizers, generation of embeddings and the training of models like BERT. However, what was most valuable was the presence of mentors which served to both answer the questions that I had and also cultivate the passion for NLP and other fields.
What were some of the specific challenges you faced during the development of the “Treatment Plan” module, and how did you overcome them?
The initial challenge we encountered was the presence of unlabelled data. Although EM2AI provided us with a substantial amount of high-quality data, manual labelling was not feasible. By creating a prototype draft of labels and then refining it with the help of dental-trained staff, we were able to expedite the labelling process. The second challenge involved the subpar performance of the native model, which stemmed from the use of a generic, off-the-shelf embedding model. To address this, we developed a dental-specific embedding model from the ground up, which significantly enhanced our performance.
What the project proves
The development and deployment of the “Treatment Plan” module again shows the proven capability development methodologies of the AIAP and 100E/SIP. It also highlights the importance of collaboration between AI engineers and domain experts in creating solutions that address real-world challenges.
Benjamin’s AI journey from a non-computing background to a skilled AI professional shows how passion, dedication, and the right opportunities can lead to a fulfilling career in AI. It also underscores the importance of programs like AIAP in nurturing diverse talent and building a thriving AI ecosystem.
Read the press release here: Q&M Dental Group Advanced “Treatment Plan” AI-Module Released Today (acnnewswire.com)