LG CHEM has developed an artificial intelligence (AI) model to predict growth after treatment with growth hormone. The company plans to develop this as a physician-specific AI platform for use in medical settings.
LG CHEM announced on the 28th that it presented a poster on 'Development of an AI-based growth prediction model' at the European Society of Pediatric Endocrinology and the European Endocrine Society congress held in Copenhagen, Denmark. This research was conducted with the participation of Professor Shim Young-seok from Ajou University Hospital and LG CHEM's DX team leader Jeong Ji-yeon.
Professor Shim Young-seok from Ajou University, who was one of the presenters, noted, 'The demand in clinical settings to estimate the effectiveness of growth hormone treatment for patients with short stature is continuously increasing,' and added, 'We have advanced the growth prediction AI model using the large-scale treatment data accumulated through LG CHEM's U-trophin long-term safety study (LG Growth Study).'
LG CHEM has built an ensemble AI model that combines and reconfigures several existing deep learning models. By training on treatment data from 3,045 short stature patients (Training Dataset), it enhanced prediction performance and used the actual growth data of 550 patients as AI performance validation data to assess prediction stability over 1 to 3 years of treatment. Conditions such as growth hormone deficiency, idiopathic short stature, inappropriate low birth weight, and Turner syndrome in girls fall under short stature diseases.
The evaluation of the AI-based model's performance revealed that the AI model predicted more accurately compared to traditional statistical modeling approaches. Notably, the AI-based model showed an average prediction error of 1.95 cm for the growth value at 1 year of treatment, based solely on initial clinical measurements (height, weight, growth hormone dosage, etc.).
LG CHEM plans to further stabilize prediction performance to ensure that it becomes a useful clinical tool for presenting changes in height percentile in relation to growth hormone treatment, promoting its application in medical practice.
Yoon Soo-young, who is responsible for life sciences and management innovation at LG CHEM, said, 'Developing a growth prediction model specialized for Korean children based on data from domestic short stature patients is highly meaningful,' and added, 'We expect that LG CHEM's innovative achievements will act as a compass for treatment in actual clinical settings.'
LG CHEM has been conducting the 'LG Growth Study (LGS)' since 2012 to secure long-term dosing data of growth hormone specifically for Korean children, making it the first in the country. LGS is a large-scale project recruiting 10,000 patients to observe long-term safety and efficacy until 2032, spanning over 20 years.