Using Digital Transformation in oral healthcare to achieve 85 percent accuracy in predictive models

Background Information

Oral healthcare is one of the most crucial parts of one’s daily life, affecting many aspects of individual well-being. Traditional methodologies lack the ability to predict intraoral forces, and since it’s also difficult to get feedback from patients, digitalizing the processes of oral scanning, tooth shape/position identification and orthodontic digital model generation is a significant opportunity. To foster better treatment and outcomes, iLittleMiracle (iLM) employs Simcenter™ Radioss® software and the Rapidminer® portfolio, both part of the Siemens Xcelerator business platform of software, hardware and services. This enables orthodontists to make accurate predictions quickly thanks to the tools’ unparalleled simulation and visualization capabilities.

About the customer

iLM was founded in 2011 and is headquartered in Beijing, China. The company is a leading brand that focuses on the early correction of malocclusion and deformity in children between 5 and 12 years old. Using digital innovation, iLM is committed to promoting the continuous innovation and popularization of medical technology for full-cycle facial development management in children.

Their challenge

Orthodontic treatment is one of the biggest categories in dentistry, comprising the highest number of cases each year. As cases increase, it’s crucial to address the shortcomings of traditional, non-digitialized methodologies.

Our solution

Using Simcenter Radioss, iLM achieved an accuracy of more than 85 percent, heightened the customer experience and enhanced analysis and prediction capabilities in treatment outcomes. The iLM research and development (R&D) process – which integrates artificial intelligence (AI), measurement data, a Digital Twin virtual jaw model and biomechanics simulation analysis – helps in the visualization of the movement and forces that act inside the oral cavity. Using Simcenter Radioss enables the doctor to perform deformity correction simulation and uses simulation to restore physical phenomena. It can perform oral 3D scanning modeling for each patient and align with customized correction plans to determine the impact of orthodontic appliances on patients’ teeth, gums and oral cavity in real time.

Within Rapidminer, iLM used a black-box model to build prediction models. The Rapidminer simulator function helped doctors better understand the causes of patient complaints and how much the most important attributes contributed to those complaints. Moreover, simulation data is used to develop predictive AI models that drive product design improvements and facilitate performance prediction.

Finally, after analysis, data can be used to predict treatment outcomes. At iLM, the Digital Twin not only provides remote visualization, it also enhances orthodontic control and fosters fast feedback. Lastly, the self-service data preparation tools in the Rapidminer platform play a vital role in enhancing iLM’s capabilities. Since the templates for patients’ medical records are in PDF format, there have been many different versions of the templates over the years. Thanks to the Rapidminer platform’s flexible template customization functions, it can extract data from different versions and years of PDF files. By employing defined extraction templates for precise data capture and batch processing to eliminate one-by-one processing and integrating with diverse data sources, Rapidminer delivers high data quality standards.

Top row: Frontal view, Lateral view
Bottom row: Upper jaw, Lower jaw
Biomechanical simulation offers various jaw views for enhanced visualization.

Results

Siemens’ unique tools and capabilities drove improvements not only in the areas of prediction and analysis, but also in patients’ comfort with dental appliances. The tools also automatically generated a table of key factors, ranking the positive or negative influence of independent variables on customer complaints from largest to smallest. This enabled quick and clear understanding. The combination of simulation and AI enables users to use data and create predictive models. The models have millisecond response times, low operation and maintenance costs, fast iterations and have highly predictive accuracy based on simulation results.

The simulator feature in Rapidminer can automatically generate a table of important factors, listing the positive or negative influence of independent variables on customer complaints in order of largest to smallest.