Glucose forecasting using genetic programming and latent glucose variability features

Objective: Characterizing blood glucose curves and providing precise patient level risk assessment of hyperglycemia using extreme value statistics and comparing these assessments with traditional indicators of glycemic variability which are not designed to specifically capture the risk of hyperglycemia. Research Design and Methods: One year return level (blood glucose level exceeded exactly once every year on average) and probability of exceeding and expected time spent above certain thresholds (600 and 400 mg/dL) per year were calculated. As a comparison, traditional metrics for glycemic variability were determined too. The effect of body mass index on extremes was also investigated using non-stationary models. Metrics were calculated on a dataset containing 170.8 patient-years of measurements of 226 patients. Results: Nine high-risk patients were identified with the novel metrics: their estimated time spent above 600 mg/dL per year were above 2 hours. These patients were at moderate risk according to the traditional metrics. Higher body mass index was associated with more extreme glucose levels. Conclusions: Through these estimates it is possible to assess each patient’s individual clinical risk of hyperglycemia even beyond the observed blood glucose levels and detection limits. Additionally, it allows the assessment of the impact of clinical characteristics and treatments on blood glucose control in a novel, mathematically well-founded and potentially clinically more useful way than the already existing indicators.

show abstract

“…The analysis was carried out using the R statistical program package version 4.1.0 29 with ExtRemes 2.1 30 and gluvarpro 4.0 31 packages.…”

Section: Programs Usedmentioning

confidence: 99%

The Use of Extreme Value Statistics to Characterize Blood Glucose Curves and Patient Level Risk Assessment of Patients With Type I Diabetes

Szigeti

Ferenci

Kovács³

2021

J Diabetes Sci Technol

View full text Add to dashboard Cite

show abstract

“…This evolution is achieved using genetic operators (inspired by how nature does with living beings) and a őtness function that evaluates how well the program solves the problem. About BG predictions in PwD, studies of different methods can be found with a high clinical accuracy [7,8]. Models generated this way are straightforward and self-explanatory because they generate arithmetic expressions.…”

Section: Related Workmentioning

confidence: 99%

Hardware real-time individualised blood glucose predictor generator based on grammars and cartesian genetic programming

Cano,

Hidalgo,

Garnica

2024

Preprint

Self Cite

View full text Add to dashboard Cite

People with diabetes need to control their blood glucose levels to avoid live-threatening situations and long-term pathologies. Efficient blood glucose control requires estimating blood glucose ahead of time for which only heuristic estimators exist. We propose a wearable hardware implementation for blood glucose prediction using a personalized model generator. We propose a novel technique based on genetic programming for designing on-chip real-time configurable hardware design, and apply it to design a hardware implementation of a personalized model generator for blood glucose prediction over a FPGA. The technique integrates grammatical evolution, Cartesian genetic programming, with a (1+λ) evolutionary strategy. The hardware implementation accomplish the training and testing of the generated model using real-patient data. The approach is validated by training and testing two models, for 30 and 60-minute forecasting predictions, for ten real patients. Each model has been created ten times totalling 200 models evaluated. The technique achieves state-of-the-art figures of merit using the Clarke error grid analysis with 98% and 90% of the predictions in clinical acceptable regions at 30- and 60-minute horizons, respectively. The technique does not suffer from hardware limitations compared to a software implementation. It offers an adaptable solution through wearable hardware and advanced model generation techniques that enables reliable high-speed predictions with minimal errors and power consumption. For the first time, the viability of combining cartesian genetic programming with a hardware implementation for creating grammatical evolution-based blood glucose prediction is demonstrated, potentially serving as an embedded system in portable devices.

show abstract

“…Some of them are blackbox models [1], others are based on analytical models [2], and most of them provide predictions for a time horizon from 15 to 120 minutes [3]. Among them, symbolic regression (SR) [4] techniques and artificial neural networks (ANNs) obtained very good performance [5]. One of the key challenges in using ANNs for glucose prediction is the lack of interpretability of the solutions they provide, limiting their usefulness in clinical practice.…”

Section: Introductionmentioning

confidence: 99%

Learning Difference Equations With Structured Grammatical Evolution for Postprandial Glycaemia Prediction

Parra,

Joedicke,

Velasco

et al. 2024

IEEE J. Biomed. Health Inform.

View full text Add to dashboard Cite

People with diabetes must carefully monitor their blood glucose levels, especially after eating. Blood glucose management requires a proper combination of food intake and insulin boluses. Glucose prediction is vital to avoid dangerous post-meal complications in treating individuals with diabetes. Although traditional methods, and also artificial neural networks, have shown high accuracy rates, sometimes they are not suitable for developing personalised treatments by physicians due to their lack of interpretability. This study proposes a novel glucose prediction method emphasising interpretability: Interpretable Sparse Identification by Grammatical Evolution. Combined with a previous clustering stage, our approach provides finite difference equations to predict postprandial glucose levels up to two hours after meals. We divide the dataset into four-hour segments and perform clustering based on blood glucose values for the two-hour window before the meal. Prediction models are trained for each cluster for the two-hour windows after meals, allowing predictions in 15-minute steps, yielding up to eight predictions at different time horizons. Prediction safety was evaluated based on Parkes Error Grid regions. Our technique produces safe predictions through explainable expressions, avoiding zones D (0.2% average) and E (0%) and reducing predictions on zone C (6.2%). In addition, our proposal has slightly better accuracy than other techniques, including sparse identification of non-linear dynamics and artificial neural networks. The results demonstrate that our proposal provides interpretable solutions without sacrificing prediction accuracy, offering a promising approach to glucose prediction in diabetes management that balances accuracy, interpretability, and computational efficiency.

show abstract

Glucose forecasting using genetic programming and latent glucose variability features

Cited by 9 publications

References 35 publications

The Use of Extreme Value Statistics to Characterize Blood Glucose Curves and Patient Level Risk Assessment of Patients With Type I Diabetes

The Use of Extreme Value Statistics to Characterize Blood Glucose Curves and Patient Level Risk Assessment of Patients With Type I Diabetes

Hardware real-time individualised blood glucose predictor generator based on grammars and cartesian genetic programming

Learning Difference Equations With Structured Grammatical Evolution for Postprandial Glycaemia Prediction

Contact Info

Product

Resources

About