Deep Multi-Output Forecasting

Fox, Ian; Ang, Lynn; Jaiswal, Mamta; Pop‐Busui, Rodica; Wiens, Jenna

doi:10.1145/3219819.3220102

Cited by 54 publications

(63 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to this, they require parameters to be set with expert knowledge. Data-driven approaches [9,10] require little knowledge of underlying physiological mechanisms, and usually can be processed with machine learning and deep learning models. Various machine learning based models have been proposed to handle blood glucose prediction problems with varying degrees of success.…”

Section: Related Workmentioning

confidence: 99%

A neural-ODE based continuous glucose monitoring measurements forecasting approach with knowledge distillation

Xing

Cao

et al. 2023

Preprint

View full text Add to dashboard Cite

Background: With the increasing use and growing popularity of insulin pumps and continuous glucose monitoring(CGM) devices, more reliable sources of data could be obtained to lead to better glucose control, treatment decision and diabetes management. The critical challenge of diabetes management is to develop accurate algorithms that can predict future blood glucose trends and manage the amount of insulin. However, in reality, continuous glucose monitoring(CGM) data is unavailable and difficult to acquire in some scenarios: for example, most type-2 diabetic patients only measure their finger sticks which are far more sparse than CGM. Few work studies glucose prediction when such continuous glucose monitoring (CGM) data is unavailable. Therefore, in this paper, we investigated the possibility of predicting the dense continuous glucose monitoring (CGM) levels purely based on sporadic self-monitoring signals such as finger sticks. Results: We proposed a novel neural ordinary differential equation (neural ODE) based method BGKD-ODE for forecasting continuous glucose monitoring (CGM), and the fundamental idea of this method is to use neural networks to learn the underlying dynamics of blood glucose metabolism in a data-driven manner. Since most of the machine learning methods can not handle such sparse data, to alleviate the training difficulty due to the data sparsity, we integrate the expert knowledge of the existing physiological model into our model. Experiments on real-world data demonstrate the strength of BGKD-ODE in accurately predicting long-term continuous glucose monitoring (CGM) levels only based on sporadic finger sticks. Conclusion: Our proposed method BGKD-ODE could be employed to obtain better insights into foresting blood glucose trends of T1DM patients. Furthermore, in the future, this model can be extended easily to type 2 diabetes with modification on reliable field knowledge.

show abstract

Section: Related Workmentioning

confidence: 99%

A neural-ODE based continuous glucose monitoring measurements forecasting approach with knowledge distillation

Xing

Cao

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…However, this was a CGM study that collected bursts of data at 3-month intervals and only considered prediction horizons of up to 30 min. 31 While this might help to protect patients from a very serious situation, it is lacking in several ways. First, such fine-grained measurements are often not practically obtainable outside of a study setting and without using a CGM device that provides measurements every 5 min.…”

Section: Introductionmentioning

confidence: 99%

The challenge of predicting blood glucose concentration changes in patients with type I diabetes

Borle

Ryan

Greiner

2021

Health Informatics J

View full text Add to dashboard Cite

Patients with Type I Diabetes (T1D) must take insulin injections to prevent the serious long term effects of hyperglycemia. They must also be careful not to inject too much insulin because this could induce (potentially fatal) hypoglycemia. Patients therefore follow a “regimen” that determines how much insulin to inject at each time, based on various measurements. We can produce an effective regimen if we can accurately predict a patient’s future blood glucose (BG) values from his/her current features. This study explores the challenges of predicting future BG by applying a number of machine learning algorithms, as well as various data preprocessing variations (corresponding to 312 [learner, preprocessed-dataset] combinations), to a new T1D dataset that contains 29,601 entries from 47 different patients. Our most accurate predictor, a weighted ensemble of two Gaussian Process Regression models, achieved a (cross-validation) [Formula: see text] loss of 2.7 mmol/L (48.65 mg/dl). This result was unexpectedly poor given that one can obtain an [Formula: see text] of 2.9 mmol/L (52.43 mg/dl) using the naive approach of simply predicting the patient’s average BG. These results suggest that the diabetes diary data that is typically collected may be insufficient to produce accurate BG prediction models; additional data may be necessary to build accurate BG prediction models over hours.

show abstract

“…These control algorithms can be classified into: (1) open-loop controls, which require patient intervention and/or external information, such as meal or exercise announcement; and (2) closed-loop controls, which do not require the patient intervention to regulate the dosage [6] but some external information can be useful in avoiding rapid BG growth [7]. In this work we consider a restricted definition of closed-loop controller in which any information that cannot be automatically passed to the controller and requires the intervention of the user is not a closed-loop controller, a point of view shared by similar works [8,6].…”

Section: Current Methods and Systems In Diabetes Managementmentioning

confidence: 99%

“…PID use previous BG samples as feedback to determine the insulin needed to drive the desired glucose concentration in human blood. Their main disadvantage is a poor adaptation to meal disturbances [8,24] and inability to individual treatment. MPC requires a mathematical model to predict future glucose concentration using current BG, insulin delivery and meal intake; then the algorithm calculates the appropriate insulin infusion rate by minimizing the difference between estimated glucose concentration from the model and the target glucose concentration on a prediction time window [19].…”

Section: Literature Reviewmentioning

confidence: 99%

“…However, improving the prediction of supervised learning approaches implies high volumes of training labeled data and there still remains the problem of designing an appropriate controller from the predicted BG level. RL has been suggested as a more promising alternative [6,8]. RL is a branch of ML that lets the agent learn by interacting with the environment, usually an artificial patient in a simulation [6].…”

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of blood glucose level control in Type 1 diabetic patients using online and offline reinforcement learning

Viroonluecha¹

View full text Add to dashboard Cite

I would like to extend my sincere appreciation to my supervisors, Professor Esteban Egea-López and Professor José Santa, for their unwavering guidance, support, and encouragement throughout my Ph.D. journey. Despite their busy schedules, they were always there to provide insightful advice, even during leisure and holiday periods and were prompt in responding to emails within a few hours. Their dedication has been truly invaluable to me and my PhD journey. I am also grateful to Professor Pablo Pavón Mariño who invited me to pursue my PhD in the GIRTEL group at the Technical University of Cartagena.

show abstract

Deep Multi-Output Forecasting

Cited by 54 publications

References 13 publications

A neural-ODE based continuous glucose monitoring measurements forecasting approach with knowledge distillation

A neural-ODE based continuous glucose monitoring measurements forecasting approach with knowledge distillation

The challenge of predicting blood glucose concentration changes in patients with type I diabetes

Evaluation of blood glucose level control in Type 1 diabetic patients using online and offline reinforcement learning

Contact Info

Product

Resources

About