Retrospective Analysis of Traditional Chinese Medicine Diagnosis and Acupuncture Treatment of Dry Eye Disease in Singapore and Machine Learning Prediction Treatment Model of “Leisanzhen”
by JOLENE CHONG
Dry Eye Disease (DED)is a common ophthalmic ailment in middle-aged and elderly patients. It has diverse etiological factors and a complex progression. Western medicine currently lacks effective treatments, leading to a long and recurring disease course for patients. Traditional Chinese Medicine (TCM) acupuncture is widely used in clinical practice to treat this condition due to its fast onset of action, its simplicity and effectiveness, and its few adverse reactions. This study provides a detailed review and summary of relevant reports on acupuncture treatment for DED. However, a retrospective analysis of cases from Singapore using only acupuncture for 13 years has not been previously reported, making this study innovative.
This study consists of two parts. Part one is a retrospective analysis of the treatment history of DED at the Singapore Chung Hwa Medical Institution’s TCM Ophthalmology Department. Data from 262 first-visit patients who received only acupuncture between January 1, 2011, and December 31, 2023, were collected. The study summarizes the characteristics of TCM acupuncture and combines them with the local environment, etiology, treatment principles, and acupoint selection for DED patients in Singapore to provide a more effective diagnostic basis for clinical practice. Part two involved randomly dividing the 262 patient data points into a training set (183 patients) and a test set (79 patients) at a 7:3 ratio. The 19 related patient baseline characteristics were used as the original dataset for machine learning models. Five machine learning models were selected for predictive modeling: K-Nearest Neighbors (KNN), CatBoost Gradient Boosting (CDC), Support Vector Machine (SVM), Random Forest (RF), and Logistic Regression (LR). The goal was to identify the clinical features associated with the use of “Tears Three Needles” (泪三針) for treating DED.
The retrospective analysis found that DED patients in Singapore are predominantly female, with a high incidence in middle-aged and elderly individuals over 50, and most commonly in low-aged elderly people aged 60-69. The main clinical symptoms include dry eyes, gritty eyes, poor sleep, dry mouth, tearing, eye pain, and blurred vision. The primary tongue and pulse characteristics are a pale-red tongue with a thin white coating and a wiry and thin pulse. The “Five TCM Syndromes of Singapore Dry Eye” are Liver-Kidney Yin Deficiency Syndrome, Lung-Kidney Yin Deficiency Syndrome, Qi Stagnation and Blood Stasis Syndrome, Disharmony between Liver and Spleen Syndrome, and Deficiency of Body Fluid Syndrome. Among these, Liver-Kidney Yin Deficiency is the main pathogenesis of the disease and shows a significant external correlation with most clinical indicators and TCM syndrome elements. Most DED patients seeking treatment in Singapore have severe dry eye. The core acupoint combination for acupuncture is the “Ten Acupoints of Singapore Dry Eye,” which includes Taiyang, Hegu, Zusanli, Zanzhu, Sanyinjiao, “Tears Three Needles” (a local empirical acupoint), Fengchi, Chengqi, Sizhukong, and Taichong. Extra-channel acupoints, acupoints on the head, face, and neck, as well as Crossing Points and Five Transport Points, are also commonly used.
The machine learning model study examined the relationship between patient baseline characteristics and the use of the “Tears Three Needles” as a treatment method. By ranking the importance of features, the study found the prediction accuracies of the five machine learning models to be: KNN 75.95%, CDC 77.22%, SVM 72.15%, RF 69.62%, and LR 73.42%. The five most relevant influencing factors for the use of “Tears Three Needles” were identified as poor sleep, Lung-Kidney Yin Deficiency Syndrome, dry mouth, age, and a thin pulse. Clinical symptoms like “poor sleep,” “dry mouth,” and “thin pulse” had a negative impact on the prediction results, while “age” was negatively correlated with the recommendation to use “Tears Three Needles.” However, a patient having “Lung-Kidney Yin Deficiency Syndrome” had a positive impact on the prediction results. These findings are valuable for guiding the clinical application of “Tears Three Needles” in treating dry eye.
This study explores the use of machine learning to combine artificial intelligence with clinical data from acupuncture practice. The aim is to improve local acupuncture clinical outcomes, make dry eye prevention and treatment more targeted, and enhance therapeutic efficacy, thereby providing significant value.
*** The above is a simplified version of Dr. (TCM) Jolene Chong’s thesis research.
