Abstract:Abstract Blood glucose (BG) prediction is an effective approach to avoid hyper- and hypoglycemia, and achieve intelligent glucose management for patients with type 1 or serious type 2 diabetes. Recent studies have tended to adopt deep learning networks to obtain improved prediction models and more accurate prediction results, which have often required significant quantities of historical continuous glucose-monitoring (CGM) data. However, for new patients with limited historical dataset, it becomes difficult to establish an acceptable deep learning network for glucose prediction. Consequently, the goal of this study was to design a novel prediction framework with instance-based and network-based deep transfer learning for cross-subject glucose prediction based on segmented CGM time series. Taking the effects of biodiversity into consideration, dynamic time warping (DTW) was applied to determine the proper source domain dataset that shared the greatest degree of similarity for new subjects. After that, a network-based deep transfer learning method was designed with cross-domain dataset to obtain a personalized model combined with improved generalization capability. In a case study, the clinical dataset demonstrated that, with additional segmented dataset from other subjects, the proposed deep transfer learning framework achieved more accurate glucose predictions for new subjects with type 2 diabetes.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the issue of data insufficiency in blood glucose prediction for newly diagnosed type 2 diabetes patients. Specifically, traditional deep learning methods require a large amount of historical continuous glucose monitoring (CGM) data to build accurate prediction models, but new patients often lack sufficient historical data. Therefore, this paper proposes a deep transfer learning framework based on instances and networks, utilizing segmented CGM time series data from other patients to build personalized blood glucose prediction models for new patients. ### Main Contributions 1. **Proposed a deep transfer learning framework**: To build personalized blood glucose prediction models for newly diagnosed type 2 diabetes patients. 2. **Instance transfer learning based on minimum dynamic time warping (DTW) distance**: Ensures minimal differences between the extended target domain and the original target domain. 3. **Network transfer learning based on the extended target domain**: By fine-tuning the pre-trained general deep learning network, ultimately achieving a personalized prediction model. 4. **Application on clinical datasets**: The effectiveness and clinical acceptability of the framework were validated through a series of experiments. ### Method Overview 1. **Instance transfer learning**: Uses the minimum DTW distance to select data from other patients that are most similar to the new patient's data, extending the target domain. 2. **Network transfer learning**: Fine-tunes the pre-trained deep learning model on the extended target domain to obtain a personalized prediction model. 3. **Data preprocessing**: Performs Z-score normalization on the raw blood glucose data to improve data distribution and model learning speed. 4. **Evaluation metrics**: Uses root mean square error (RMSE), mean absolute error (MAE), and mean absolute relative difference (MARD) to evaluate model performance, and assesses clinical acceptability through prediction error grid analysis (PRED-EGA). ### Experimental Results - **Size of the extended target domain**: Experiments determined that when the training dataset includes data from 15 patients, the model performance tends to stabilize. - **Combined application of instance transfer and network transfer**: Experimental results show that combining instance transfer with minimum DTW distance and network transfer significantly improves the accuracy of blood glucose prediction, especially within the 30-minute and 60-minute prediction time ranges. ### Conclusion The deep transfer learning framework proposed in this study can effectively utilize limited historical data to build accurate blood glucose prediction models for newly diagnosed type 2 diabetes patients, having significant clinical application value.

Deep transfer learning: a novel glucose prediction framework for new subjects with type 2 diabetes

Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients

A Deep Transfer Learning Model for Personalized Blood Glucose Prediction

Personalized Glucose Prediction with Clustering-Based Domain Adaptation for Type 2 Diabetes

Transfer and Incremental Learning Method for Blood Glucose Prediction of New Subjects with Type 1 Diabetes

A deep learning approach to diabetic blood glucose prediction

A Novel Hybrid Deep Learning Model for Blood Glucose Prediction with Extended Prediction Horizons in Type-1 Diabetic Patients

Deep Multitask Learning by Stacked Long Short-Term Memory for Predicting Personalized Blood Glucose Concentration

Privacy Preserved Blood Glucose Level Cross-Prediction: An Asynchronous Decentralized Federated Learning Approach

A State-by-State online transfer learning framework with incremental clustering for blood glucose prediction

An autonomous channel deep learning framework for blood glucose prediction

Investigating the Impact of Multiple Inputs on Predicting Glucose Level in Individuals with Type-1 Diabetes through Deep Learning Technique

Blood Glucose Monitoring Using Non-Invasive Features of Wearable Devices and Machine Learning

A Hybrid Deep Learning Model for the Blood Glucose Prediction

A novel interpretable deep transfer learning combining diverse learnable parameters for improved T2D prediction based on single-cell gene regulatory networks

Prediction of blood glucose concentration for type 1 diabetes based on echo state networks embedded with incremental learning

A personalized multitasking framework for real-time prediction of blood glucose levels in type 1 diabetes patients

Deep learning and regression approaches to forecasting blood glucose levels for type 1 diabetes

Machine Learning and Deep Learning Models for Nocturnal High- and Low-Glucose Prediction in Adults with Type 1 Diabetes

Development of a Deep Learning Model for Dynamic Forecasting of Blood Glucose Level for Type 2 Diabetes Mellitus: Secondary Analysis of a Randomized Controlled Trial