[DETECTION, ON THE OCCIPITAL BONE, OF ELASTIC (ULTRASONIC) WAVES INDUCED BY LASER IMPULSES IN THE EYE OF A RABBIT].

L. Amar,M. Brumă,P. Desvignes,M. Leblanc,G. Perdriel,M. Velghe

1964-11-16

Comptes rendus hebdomadaires des séances de l Académie des sciences

Abstract:

What problem does this paper attempt to address?

A Randomized Clinical Trial for Meal Bolus Decision Using Learning-based Control in Adults With Type 2 Diabetes

Wei Liu,Deheng Cai,Rui Zhang,Xiuying Zhang,Xiaoling Cai,Liyuan Tao,Xueyao Han,Yingying Luo,Meng Li,Wenjing Wu,Yumin Ma,Dawei Shi,Linong Ji

DOI: https://doi.org/10.1210/clinem/dgae143

2024-03-07

The Journal of Clinical Endocrinology & Metabolism

Abstract:Abstract Context We propose an artificial-pancreas-like algorithm (AP-A) that could automatically determine the preprandial insulin dose based on intermittently scanned continuous glucose monitoring (isCGM) data trajectories in multiple dose injection (MDI) therapy. Objective We aim to determine whether preprandial insulin dose adjustments guided by the AP-A are as effective and safe as physician decisions. Methods We performed a randomized, single-blind, clinical trial at a tertiary, referral hospital in Beijing, China. Type 2 diabetes participants were eligible if they were aged 18 years or older, with a glycated hemoglobin A1c of 8.0% or higher. Eligible participants were randomly assigned (1:1) to the AP-A arm supervised by physician and the conventional physician treatment arm. The primary objective was to compare percentage time spent with sensor glucose level in 3.9 to 10.0 mmol/L (TIR) between the 2 study arms. Safety was assessed by the percentage time spent with sensor glucose level below 3.0 mmol/L (TBR). Results A total of 140 participants were screened, of whom 119 were randomly assigned to the AP-A arm (n = 59) or physician arm (n = 60). The TIR achieved by the AP-A arm was statistically noninferior compared with the control arm (72.4% [63.3%-82.1%] vs 71.2% [54.9%-81.4%]), with a median difference of 1.33% (95% CI, −6.00 to 10.94, noninferiority margin −7.5%). TBR was also statistically noninferior between the AP-A and control arms (0.0% [0.0%-0.0%] vs 0.0% [0.0%-0.0%]), respectively; median difference (95% CI, 0.00% [0.00%-0.00%], noninferiority margin 2.0%). Conclusion The AP-A–supported physician titration of preprandial insulin dosage offers noninferior glycemic control compared with optimal physician care in type 2 diabetes.

endocrinology & metabolism
Optimized Glycemic Control of Type 2 Diabetes with Reinforcement Learning: a Proof-of-concept Trial.

Guangyu Wang,Xiaohong Liu,Zhen Ying,Guoxing Yang,Zhiwei Chen,Zhiwen Liu,Min Zhang,Hongmei Yan,Yuxing Lu,Yuanxu Gao,Kanmin Xue,Xiaoying Li,Ying Chen

DOI: https://doi.org/10.1038/s41591-023-02552-9

IF: 82.9

2023-01-01

Nature Medicine

Abstract:The personalized titration and optimization of insulin regimens for treatment of type 2 diabetes (T2D) are resource-demanding healthcare tasks. Here we propose a model-based reinforcement learning (RL) framework (called RL-DITR), which learns the optimal insulin regimen by analyzing glycemic state rewards through patient model interactions. When evaluated during the development phase for managing hospitalized patients with T2D, RL-DITR achieved superior insulin titration optimization (mean absolute error (MAE) of 1.10 ± 0.03 U) compared to other deep learning models and standard clinical methods. We performed a stepwise clinical validation of the artificial intelligence system from simulation to deployment, demonstrating better performance in glycemic control in inpatients compared to junior and intermediate-level physicians through quantitative (MAE of 1.18 ± 0.09 U) and qualitative metrics from a blinded review. Additionally, we conducted a single-arm, patient-blinded, proof-of-concept feasibility trial in 16 patients with T2D. The primary outcome was difference in mean daily capillary blood glucose during the trial, which decreased from 11.1 (±3.6) to 8.6 (±2.4) mmol L −1 ( P < 0.01), meeting the pre-specified endpoint. No episodes of severe hypoglycemia or hyperglycemia with ketosis occurred. These preliminary results warrant further investigation in larger, more diverse clinical studies. ClinicalTrials.gov registration: NCT05409391 .
Intermittent Control for Safe Long-Acting Insulin Intensification for Type 2 Diabetes: In-Silico Experiment

Anas El Fathi,Mohammadreza Ganji,Dimitri Boiroux,Henrik Bengtsson,Marc D. Breton

2023-09-17

Abstract:Around a third of type 2 diabetes patients (T2D) are escalated to basal insulin injections. Basal insulin dose is titrated to achieve a tight glycemic target without undue hypoglycemic risk. In the standard of care (SoC), titration is based on intermittent fasting blood glucose (FBG) measurements. Lack of adherence and the day-to-day variabilities in FBG measurements are limiting factors to the existing insulin titration procedure. We propose an adaptive receding horizon control strategy where a glucose-insulin fasting model is identified and used to predict the optimal basal insulin dose. This algorithm is evaluated in \textit{in-silico} experiments using the new UVA virtual lab (UVlab) and a set of T2D avatars matched to clinical data (NCT01336023). Compared to SoC, we show that this control strategy can achieve the same glucose targets faster (as soon as week 8) and safer (increased hypoglycemia protection and robustness to missing FBG measurements). Specifically, when insulin is titrated daily, a time-in-range (TIR, 70--180 mg/dL) of 71.4$\pm$20.0\% can be achieved at week eight and maintained at week 52 (72.6$\pm$19.6%) without an increased hypoglycemia risk as measured by time under 70 mg/dL (TBR, week 8: 1.3$\pm$1.9% and week 52: 1.2$\pm$1.9%), when compared to the SoC (TIR at week 8: 59.3$\pm$28.0% and week:52 72.1$\pm$22.3%, TBR at week 8: 0.5$\pm$1.3% and week 52: 2.8$\pm$3.4%). Such an approach can potentially reduce treatment inertia and prescription complexity, resulting in improved glycemic outcomes for T2D using basal insulin injections.

Systems and Control
Real-time Artificial Intelligence Assisted Insulin Dosage Titration System for Glucose Control in Type 2 Diabetic Patients: a Proof of Concept Study

Chen Ying,Chen Zhiwei,Zhao Lin,Li Simin,Ying Zhen,Yu Peng,Yan Hongmei,Chen Hong,Yang Chun,Zhang Jiyang,Meng Qingnan,Liu Yuchen,Cao Ling,Shen Yanting,Hu Chunyan,Huang Huiqun,Li Xiaomu,Bian Hua,Li Xiaoying

DOI: https://doi.org/10.1007/s44194-023-00020-7

2023-01-01

Current Medicine

Abstract:Objective This study aims to develop an insulin dosage adjustment model using machine learning of high quality electronic health records (EHRs) notes and then to form an artificial intelligence-based insulin clinical decision support workflow (iNCDSS) implemented in the HIS system to give a real-time recommendation of insulin dosage titration. The efficacy and safety in clinical practice is evaluated in this proof-of-concept study. Research design and methods We extracted patient-specific and time-varying features from the original EHRs data and performed machine learning analysis through 5-fold cross validation. In the patient-blind, single-arm interventional study, insulin dosage was titrated according to iNCDSS in type 2 diabetic inpatients for up to 7 d or until hospital discharge. The primary end point of the trial was the difference in glycemic control as measured by mean daily blood glucose concentration during the intervention period. Results A total of 3275 type 2 diabetic patients with 38,406 insulin counts were included for the model analysis. The XGBoost model presented the best performance with root mean square error (RMSE) of 1.06 unit and mean absolute relative difference (MARD) of 6.0% in the training dataset, and RMSE of 1.30 unit and MARD of 6.9% in the testing dataset. Twenty-three patients with T2DM (male 14, 60.9%; age 58.8 ± 10.7 years; duration of diabetes 11.8 ± 8.8 years, HbA1c 9.1 ± 1.1%) were enrolled in the proof of concept trial. The duration of iNCDSS intervention was 7.0 ± 0.1 d. The insulin dose recommended by iNCDSS was accepted by physicians in 97.8%. The mean daily capillary blood glucose was markedly improved during the intervention period, with a reduction of mean daily capillary BG from 11.3(8.0, 13.9) mmol/L in the first 24 h to 7.9(6.5,8.9) mmol/L in the last 24 h of the trial ( P < 0.001). In addition, the time range below 3.9 mmol/L was decreased from 1.1% to 0.5%. Conclusions The clinical decision support system of insulin dosage titration developed using a machine learning algorithm based on the EHRs data was effective and safe in glycemic control in in type 2 diabetic inpatients. Trial registrations ClinicalTrials.gov Identifier: NCT04053959 .
Attention Networks for Personalized Mealtime Insulin Dosing in People with Type 1 Diabetes

Anas El Fathi,Elliott Pryor,Marc D. Breton

2024-06-19

Abstract:Calculating mealtime insulin doses poses a significant challenge for individuals with Type 1 Diabetes (T1D). Doses should perfectly compensate for expected post-meal glucose excursions, requiring a profound understanding of the individual's insulin sensitivity and the meal macronutrients'. Usually, people rely on intuition and experience to develop this understanding. In this work, we demonstrate how a reinforcement learning agent, employing a self-attention encoder network, can effectively mimic and enhance this intuitive process. Trained on 80 virtual subjects from the FDA-approved UVA/Padova T1D adult cohort and tested on twenty, self-attention demonstrates superior performance compared to other network architectures. Results reveal a significant reduction in glycemic risk, from 16.5 to 9.6 in scenarios using sensor-augmented pump and from 9.1 to 6.7 in scenarios using automated insulin delivery. This new paradigm bypasses conventional therapy parameters, offering the potential to simplify treatment and promising improved quality of life and glycemic outcomes for people with T1D.

Quantitative Methods,Artificial Intelligence
Insulin dose optimization using an automated artificial intelligence-based decision support system in youths with type 1 diabetes

Revital Nimri,Tadej Battelino,Lori M. Laffel,Robert H. Slover,Desmond Schatz,Stuart A. Weinzimer,Klemen Dovc,Thomas Danne,Moshe Phillip,Shlomit Shalitin,Rachel Bello,Michal Nevo-Shenker,Naama Fisch-Shvalb,Galit Shiovitch-Mantzuri,Orit Choresh,Irit Drutz,Yehiel Nava,Alona Hemo,Orna Hermon,Rachel Nave,Natasa Bratina,Darja Smigoe-Schweiger,Brigita Mali,Ana Gianini,Urska Sever,Barbara Murn Berkopec,Michelle Katz,Elvira Isganaitis,Sanjeev Mehta,Heidi Quinn,Nisha Naik,Zijing Guo,Lisa Volkening,Gregory Forlenza,R. Paul Wadwa,G. Todd Alonso,Laurel Messer,Lindsey Towers,Katie Thivener,Cari Berget,Samantha Lange,Emily Jost,Maria Rossick-Solis,Michael Haller,Paul Hiers,Laura Jacobsen,Madison Smith,Anastasia O’Neill,Jennifer Hosford,Alexis Perry,Eda Cengiz,Jennifer Sherr,Kathryn Gibbons,Lori Carria,Melinda Zgorski,Torben Biester,Olga Kordonouri,Thekla von dem Berge,Sarah Biester,Kerstin Remus,

DOI: https://doi.org/10.1038/s41591-020-1045-7

IF: 82.9

2020-09-01

Nature Medicine

Abstract:Despite the increasing adoption of insulin pumps and continuous glucose monitoring devices, most people with type 1 diabetes do not achieve their glycemic goals<a href="/articles/s41591-020-1045-7#ref-CR1">1</a>. This could be related to a lack of expertise or inadequate time for clinicians to analyze complex sensor-augmented pump data. We tested whether frequent insulin dose adjustments guided by an automated artificial intelligence-based decision support system (AI-DSS) is as effective and safe as those guided by physicians in controlling glucose levels. ADVICE4U was a six-month, multicenter, multinational, parallel, randomized controlled, non-inferiority trial in 108 participants with type 1 diabetes, aged 10–21 years and using insulin pump therapy (ClinicalTrials.gov no. NCT03003806). Participants were randomized 1:1 to receive remote insulin dose adjustment every three weeks guided by either an AI-DSS, (AI-DSS arm, n = 54) or by physicians (physician arm, n = 54). The results for the primary efficacy measure—the percentage of time spent within the target glucose range (70–180 mg dl−1 (3.9–10.0 mmol l−1))—in the AI-DSS arm were statistically non-inferior to those in the physician arm (50.2 ± 11.1% versus 51.6 ± 11.3%, respectively, P < 1 × 10−7). The percentage of readings below 54 mg dl−1 (<3.0 mmol l−1) within the AI-DSS arm was statistically non-inferior to that in the physician arm (1.3 ± 1.4% versus 1.0 ± 0.9%, respectively, P < 0.0001). Three severe adverse events related to diabetes (two severe hypoglycemia, one diabetic ketoacidosis) were reported in the physician arm and none in the AI-DSS arm. In conclusion, use of an automated decision support tool for optimizing insulin pump settings was non-inferior to intensive insulin titration provided by physicians from specialized academic diabetes centers.

biochemistry & molecular biology,cell biology,medicine, research & experimental
Basal Glucose Control in Type 1 Diabetes Using Deep Reinforcement Learning: An In Silico Validation

Taiyu Zhu,Kezhi Li,Pau Herrero,Pantelis Georgiou

DOI: https://doi.org/10.1109/jbhi.2020.3014556

IF: 7.7

2021-04-01

IEEE Journal of Biomedical and Health Informatics

Abstract:People with Type 1 diabetes (T1D) require regular exogenous infusion of insulin to maintain their blood glucose concentration in a therapeutically adequate target range. Although the artificial pancreas and continuous glucose monitoring have been proven to be effective in achieving closed-loop control, significant challenges still remain due to the high complexity of glucose dynamics and limitations in the technology. In this work, we propose a novel deep reinforcement learning model for single-hormone (insulin) and dual-hormone (insulin and glucagon) delivery. In particular, the delivery strategies are developed by double Q-learning with dilated recurrent neural networks. For designing and testing purposes, the FDA-accepted UVA/Padova Type 1 simulator was employed. First, we performed long-term generalized training to obtain a population model. Then, this model was personalized with a small data-set of subject-specific data. In silico results show that the single and dual-hormone delivery strategies achieve good glucose control when compared to a standard basal-bolus therapy with low-glucose insulin suspension. Specifically, in the adult cohort (n = 10), percentage time in target range 70, 180 mg/dL improved from 77.6% to 80.9% with single-hormone control, and to 85.6% with dual-hormone control. In the adolescent cohort (n = 10), percentage time in target range improved from 55.5% to [Formula: see text] with single-hormone control, and to 78.8% with dual-hormone control. In all scenarios, a significant decrease in hypoglycemia was observed. These results show that the use of deep reinforcement learning is a viable approach for closed-loop glucose control in T1D.

computer science, interdisciplinary applications,mathematical & computational biology,medical informatics, information systems
Managing the diversification risk.

D. Gregory,K. Kaufman

1983-12-01

Abstract:
Managing antibiotic associated diarrhoea

F. Barbut,J. Meynard

DOI: https://doi.org/10.1136/bmj.324.7350.1345

2002-06-08

British Medical Journal

Abstract:Papers p 1361 Diarrhoea is a common adverse effect of antibiotic treatments. Antibiotic associated diarrhoea occurs in about 5-30% of patients either early during antibiotic therapy or up to two months after the end of the treatment.1–3 The frequency of antibiotic associated diarrhoea depends on the definition of diarrhoea, the inciting antimicrobial agents, and host factors. Almost all antibiotics, particularly those that act on anaerobes, can cause diarrhoea, but the risk is higher with aminopenicillins, a combination of aminopenicillins and clavulanate, cephalosporins, and clindamycin. 1 4 5 Host factors for antibiotic associated diarrhoea include age over 65, immunosuppression, being in an intensive care unit, and prolonged hospitalisation.6 Clinical presentations of antibiotic associated diarrhoea range from mild diarrhoea to fulminant pseudomembranous colitis. The latter is characterised by a watery diarrhoea, fever (in 80% of cases), leucocytosis (80%), and the presence of pseudomembranes on endoscopic examination. Severe complications include toxic megacolon, perforation, and shock. Antibiotic associated diarrhoea results from disruption of the normal microflora of the gut by antibiotics. This microflora, composed of 1011 bacteria per gram of intestinal content, forms …
Use of a decision support tool and quick start onboarding tool in individuals with type 1 diabetes using advanced automated insulin delivery: a single-arm multi-phase intervention study

Shekhar Sehgal,Martin De Bock,Benyamin Grosman,Jonathan Williman,Natalie Kurtz,Vanessa Guzman,Andrea Benedetti,Anirban Roy,Kamuran Turksoy,Magaly Juarez,Shirley Jones,Carla Frewen,Antony Watson,Barry Taylor,Benjamin J Wheeler

DOI: https://doi.org/10.1186/s12902-024-01709-y

2024-08-30

Abstract:Background: Multiple clinician adjustable parameters impact upon glycemia in people with type 1 diabetes (T1D) using Medtronic Mini Med 780G (MM780G) AHCL. These include glucose targets, carbohydrate ratios (CR), and active insulin time (AIT). Algorithm-based decision support advising upon potential settings adjustments may enhance clinical decision-making. Methods: Single-arm, two-phase exploratory study developing decision support to commence and sustain AHCL. Participants commenced investigational MM780G, then 8 weeks Phase 1-initial optimization tool evaluation, involving algorithm-based decision support with weekly AIT and CR recommendations. Clinicians approved or rejected CR and AIT recommendations based on perceived safety per protocol. Co-design resulted in a refined algorithm evaluated in a further identically configured Phase 2. Phase 2 participants also transitioned to commercial MM780G following "Quick Start" (algorithm-derived tool determining initial AHCL settings using daily insulin dose and weight). We assessed efficacy, safety, and acceptability of decision support using glycemic metrics, and the proportion of accepted CR and AIT settings per phase. Results: Fifty three participants commenced Phase 1 (mean age 24.4; Hba1c 61.5mmol/7.7%). The proportion of CR and AIT accepted by clinicians increased between Phases 1 and 2 respectively: CR 89.2% vs. 98.6%, p < 0.01; AIT 95.2% vs. 99.3%, p < 0.01. Between Phases, mean glucose percentage time < 3.9mmol (< 70mg/dl) reduced (2.1% vs. 1.4%, p = 0.04); change in mean TIR 3.9-10mmol/L (70-180mg/dl) was not statistically significant: 72.9% ± 7.8 and 73.5% ± 8.6. Quick start resulted in stable TIR, and glycemic metrics compared to international guidelines. Conclusion: The co-designed decision support tools were able to deliver safe and effective therapy. They can potentially reduce the burden of diabetes management related decision making for both health care practitioners and patients. Trial registration: Prospectively registered with Australia/New Zealand Clinical Trials Registry(ANZCTR) on 30th March 2021 as study ACTRN12621000360819.
Reinforcement Learning-Based Adaptive Insulin Advisor for Individuals with Type 1 Diabetes Patients under Multiple Daily Injections Therapy

Qingnan Sun,Marko V. Jankovic,Stavroula G. Mougiakakou

DOI: https://doi.org/10.1109/embc.2019.8857178

2019-07-01

Abstract:The existing adaptive basal-bolus advisor (ABBA) was further developed to benefit patients under insulin therapy with multiple daily injections (MDI). Three different in silico experiments were conducted with the DMMS.R simulator to validate the approach of combined use of self-monitoring of blood glucose (SMBG) and insulin injection devices, e.g. insulin pen, as are used by the majority of type 1 diabetes patients under insulin therapy. The proposed approach outperforms the conventional method, as it increases the time spent within the target range and simultaneously reduces the risks of hyperglycaemic and hypoglycaemic events.
Simple meal announcements and pramlintide delivery versus carbohydrate counting in type 1 diabetes with automated fast-acting insulin aspart delivery: a randomised crossover trial in Montreal, Canada

Elisa Cohen,Michael A Tsoukas,Laurent Legault,Michael Vallis,Julia E Von Oettingen,Emilie Palisaitis,Madison Odabassian,Jean-François Yale,Natasha Garfield,Nikita Gouchie-Provencher,Joanna Rutkowski,Adnan Jafar,Milad Ghanbari,Ahmad Haidar

DOI: https://doi.org/10.1016/s2589-7500(24)00092-x

IF: 36.615

2024-06-21

The Lancet Digital Health

Abstract:Summary Background In type 1 diabetes, carbohydrate counting is the standard of care to determine prandial insulin needs, but it can negatively affect quality of life. We developed a novel insulin-and-pramlintide closed-loop system that replaces carbohydrate counting with simple meal announcements. Methods We performed a randomised crossover trial assessing 14 days of (1) insulin-and-pramlintide closed-loop system with simple meal announcements, (2) insulin-and-placebo closed-loop system with carbohydrate counting, and (3) insulin-and-placebo closed-loop system with simple meal announcements. Participants were recruited at McGill University Health Centre (Montreal, QC, Canada). Eligible participants were adults (aged ≥18 years) and adolescents (aged 12–17 years) with type 1 diabetes for at least 1 year. Participants were randomly assigned in a 1:1:1:1:1:1 ratio to a sequence of the three interventions, with faster insulin aspart used in all interventions. Each intervention was separated by a 14–45-day wash-out period, during which participants reverted to their usual insulin. During simple meal announcement interventions, participants triggered a prandial bolus at mealtimes based on a programmed fixed meal size, whereas during carbohydrate counting interventions, participants manually entered the carbohydrate content of the meal and an algorithm calculated the prandial bolus based on insulin-to-carbohydrate ratio. Two primary comparisons were predefined: the percentage of time in range (glucose 3·9–10·0 mmol/L) with a non-inferiority margin of 6·25% (non-inferiority comparison); and the mean Emotional Burden subscale score of the Diabetes Distress Scale (superiority comparison), comparing the insulin-and-placebo system with carbohydrate counting minus the insulin-and-pramlintide system with simple meal announcements. Analyses were performed on a modified intention-to-treat basis, excluding participants who did not complete all interventions. Serious adverse events were assessed in all participants. This trial is registered on ClinicalTrials.gov , NCT04163874 . Findings 32 participants were enrolled between Feb 14, 2020, and Oct 5, 2021; two participants withdrew before study completion. 30 participants were analysed, including 15 adults (nine female, mean age 39·4 years [SD 13·8]) and 15 adolescents (eight female, mean age 15·7 years [1·3]). Non-inferiority of the insulin-and-pramlintide system with simple meal announcements relative to the insulin-and-placebo system with carbohydrate counting was reached (difference –5% [95% CI –9·0 to –0·7], non-inferiority p<0·0001). No statistically significant difference was found in the mean Emotional Burden score between the insulin-and-pramlintide system with simple meal announcements and the insulin-and-placebo system with carbohydrate counting (difference 0·01 [SD 0·82], p=0·93). With the insulin-and-pramlintide system with simple meal announcements, 14 (47%) participants reported mild gastrointestinal symptoms and two (7%) reported moderate symptoms, compared with two (7%) participants reporting mild gastrointestinal symptoms on the insulin-and-placebo system with carbohydrate counting. No serious adverse events occurred. Interpretation The insulin-and-pramlintide system with simple meal announcements alleviated carbohydrate counting without degrading glucose control, although quality of life as measured by the Emotional Burden score was not improved. Longer and larger studies with this novel approach are warranted. Funding Juvenile Diabetes Research Foundation.

medicine, general & internal,medical informatics
Simulation-driven optimization of insulin therapy profiles in a commercial hybrid closed-loop system

Jenny L Diaz C,María F Villa-Tamayo,Marcela Moscoso-Vasquez,Patricio Colmegna

DOI: https://doi.org/10.1016/j.cmpb.2023.107830

Abstract:Background: Automated insulin delivery (AID) has represented a breakthrough in managing type 1 diabetes (T1D), showing safe and effective glucose control extensively across the board. However, metabolic variability still poses a challenge to commercial hybrid closed-loop (HCL) solutions, whose performance depends on customizable insulin therapy profiles. In this work, we propose an Identification-Replay-Optimization (IRO) approach to optimize gradually and safely such profiles for the Control-IQ AID algorithm. Methods: Closed-loop data are generated using the full adult cohort of the UVA/Padova T1D simulation platform in diverse glycemic scenarios. For each subject, daily records are processed and used to estimate a personalized model of the underlying insulin-glucose dynamics. Every two weeks, all identified models are integrated into an optimization procedure where daily basal and bolus profiles are adjusted so as to minimize the risks for hypo- and hyperglycemia. The proposed strategy is tested under different scenarios of metabolic and behavioral variability in order to evaluate the efficacy and convergence of the proposed strategy. Finally, glycemic metrics between cycles are compared using paired t-tests with p<0.05 as the significance threshold. Results: Simulations reveal that the proposed IRO approach was able to improve glucose control over time by safely mitigating the risks for both hypo- and hyperglycemia. Furthermore, smaller changes were recommended at each cycle, indicating convergence when simulation conditions were maintained. Conclusions: The use of reliable simulation-driven tools capable of accurately reproducing field-collected data and predicting changes can substantially shorten the process of optimizing insulin therapy, adjusting it to metabolic changes and leading to improved glucose control.
Performance of an Automated Insulin Delivery System: Results of Early Phase Feasibility Studies

Mark Christiansen,Amy Bartee,Amy Lalonde,Richard E Jones,Michelle Katz,Howard Wolpert,Ronald Brazg

DOI: https://doi.org/10.1089/dia.2020.0318

Abstract:Background: Automated insulin delivery (AID) systems have demonstrated improvements in time-in-range (TIR, blood glucose 70-180 mg/dL) without increasing hypoglycemia. Testing a closed-loop system in an inpatient environment with supervised challenges allows for initial evaluation of performance and safety of the system. Methods: Adults with type 1 diabetes (T1D) were enrolled into two similar studies (n = 10 per study), with 3-day inpatient analysis periods. Participants tested a Lilly hybrid closed-loop (HCL) system comprising an investigational insulin pump, insulin lispro, a pump-embedded model predictive control algorithm, a continuous glucose monitor (CGM), and an external dedicated controller. Each protocol included meal-related and exercise challenges to simulate real-world diabetes self-management errors. Only study staff interacted with the HCL system. Performance was assessed using standard CGM metrics overall and within prespecified periods. Results: Participants (25% male) had mean ± standard deviation (SD) age 44.7 ± 14.2 years, T1D duration 30.2 ± 11.1 years, A1C 7.2% ± 0.8%, and insulin usage 0.53 ± 0.21 U/(kg·day). Percentage TIR 70-180 mg/dL (mean ± SD) was 81.2 ± 8.4 overall, 85.2 ± 8.1 outside of challenge periods, 97.3 ± 5.3 during the nocturnal periods, and 74.5 ± 16.2 for the postprandial periods. During challenge periods, percentage TIR for the overbolus challenge was 65.4 ± 29.2 and that for the delayed bolus challenge was 57.1 ± 25.1. No adverse events (AEs), serious AEs, or unanticipated adverse device events occurred while participants were using the HCL system. Conclusions: In participants with T1D, Lilly AID system demonstrated expected algorithm performance and safety with satisfactory glycemic outcomes overall and in response to simulated diabetes management challenges. Additional studies in less supervised conditions and with broader patient populations are warranted. ClinicalTrials.gov Registration number NCT03743285, NCT03849612.
Reinforcement Learning: A Paradigm Shift in Personalized Blood Glucose Management for Diabetes

Lehel Dénes-Fazakas,László Szilágyi,Levente Kovács,Andrea De De Gaetano,György Eigner

DOI: https://doi.org/10.3390/biomedicines12092143

IF: 4.757

2024-09-22

Biomedicines

Abstract:Background/Objectives: Managing blood glucose levels effectively remains a significant challenge for individuals with diabetes. Traditional methods often lack the flexibility needed for personalized care. This study explores the potential of reinforcement learning-based approaches, which mimic human learning and adapt strategies through ongoing interactions, in creating dynamic and personalized blood glucose management plans. Methods: We developed a mathematical model specifically for patients with type IVP diabetes, validated with data from 10 patients and 17 key parameters. The model includes continuous glucose monitoring (CGM) noise and random carbohydrate intake to simulate real-life conditions. A closed-loop system was designed to enable the application of reinforcement learning algorithms. Results: By implementing a Policy Optimization (PPO) branch, we achieved an average Time in Range (TIR) metric of 73%, indicating improved blood glucose control. Conclusions: This study presents a personalized insulin therapy solution using reinforcement learning. Our closed-loop model offers a promising approach for improving blood glucose regulation, with potential applications in personalized diabetes management.

biochemistry & molecular biology,medicine, research & experimental,pharmacology & pharmacy
Real-Life Use of Automated Insulin Delivery in Individuals With Type 2 Diabetes

Chiara Fabris,Boris Kovatchev

DOI: https://doi.org/10.1177/19322968241274786

2024-08-24

Abstract:Background: The objective of this work is to document performance of automated insulin delivery (AID) during real-life use in type 2 diabetes (T2D). Methods: A retrospective analysis was performed of continuous glucose monitoring and insulin delivery data from 796 individuals with T2D, who transitioned from 1-month predictive low-glucose suspend (PLGS) use to 3-month AID use, in real-life settings. Primary outcome was change of time in range (TIR = 70-180 mg/dL) from PLGS to AID. Secondary outcomes included time above/below range (TAR/TBR) and total daily insulin (TDI). Results: Compared with PLGS, AID increased TIR on average from 63.2% to 72.6%, decreased TAR from 36.2% to 26.8%, and increased TDI from 70.2 to 76.3 U (all P < .001), without significant change to TBR. Glycemic improvements were more pronounced in those with worse glycemic control during PLGS use (P < .001). Conclusions: Real-life use of AID led to a rapid and sustained improvement of glycemic control in individuals with T2D.
A Dual Mode Adaptive Basal-Bolus Advisor Based on Reinforcement Learning

Qingnan Sun,Marko V. Jankovic,Joo Budzinski,Brett Moore,Peter Diem,Christoph Stettler,Stavroula G. Mougiakakou,Joao Budzinski

DOI: https://doi.org/10.1109/jbhi.2018.2887067

IF: 7.7

2019-11-01

IEEE Journal of Biomedical and Health Informatics

Abstract:Self-monitoring of blood glucose (SMBG) and continuous glucose monitoring (CGM) are commonly used by type 1 diabetes (T1D) patients to measure glucose concentrations. The proposed adaptive basal-bolus algorithm (ABBA) supports inputs from either SMBG or CGM devices to provide personalised suggestions for the daily basal rate and prandial insulin doses on the basis of the patients' glucose level on the previous day. The ABBA is based on reinforcement learning, a type of artificial intelligence, and was validated in silico with an FDA-accepted population of 100 adults under different realistic scenarios lasting three simulated months. The scenarios involve three main meals and one bedtime snack per day, along with different variabilities and uncertainties for insulin sensitivity, mealtime, carbohydrate amount, and glucose measurement time. The results indicate that the proposed approach achieves comparable performance with CGM or SMBG as input signals, without influencing the total daily insulin dose. The results are a promising indication that AI algorithmic approaches can provide personalised adaptive insulin optimization and achieve glucose control-independent of the type of glucose monitoring technology.
Estimating a Personalized Basal Insulin Dose from Short-Term Closed-Loop Data in Type 2 Diabetes

Sarah Ellinor Engell,Tinna Björk Aradóttir,Tobias K. S. Ritschel,Henrik Bengtsson,John Bagterp Jørgensen

DOI: https://doi.org/10.48550/arXiv.2211.14149

2022-11-25

Systems and Control

Abstract:In type 2 diabetes (T2D) treatment, finding a safe and effective basal insulin dose is a challenge. The dose-response is highly individual and to ensure safety, people with T2D titrate by slowly increasing the daily insulin dose to meet treatment targets. This titration can take months. To ease and accelerate the process, we use short-term artificial pancreas (AP) treatment tailored for initial titration and apply it as a diagnostic tool. Specifically, we present a method to automatically estimate a personalized daily dose of basal insulin from closed-loop data collected with an AP. Based on AP-data from a stochastic simulation model, we employ the continuous-discrete extended Kalman filter and a maximum likelihood approach to estimate parameters in a simple dose-response model for 100 virtual people. With the identified model, we compute a daily dose of basal insulin to meet treatment targets for each individual. We test the personalized dose and evaluate the treatment outcomes against clinical reference values. In the tested simulation setup, the proposed method is feasible. However, more extensive tests will reveal whether it can be deemed safe for clinical implementation.
Challenging common bolus advisor for self-monitoring type-I diabetes patients using Reinforcement Learning

Frédéric Logé,Erwan Le Pennec,Habiboulaye Amadou-Boubacar

DOI: https://doi.org/10.48550/arXiv.2007.11880

2020-07-23

Abstract:Patients with diabetes who are self-monitoring have to decide right before each meal how much insulin they should take. A standard bolus advisor exists, but has never actually been proven to be optimal in any sense. We challenged this rule applying Reinforcement Learning techniques on data simulated with T1DM, an FDA-approved simulator developed by Kovatchev et al. modeling the gluco-insulin interaction. Results show that the optimal bolus rule is fairly different from the standard bolus advisor, and if followed can actually avoid hypoglycemia episodes.

Machine Learning
In children and young people with type 1 diabetes using Pump therapy, an additional 40% of the insulin dose for a high‐fat, high‐protein breakfast improves postprandial glycaemic excursions: A cross‐over trial

Tenele A. Smith,Carmel E. Smart,Michelle E. J. Fuery,Peter P. Howley,Brigid A. Knight,Mark Harris,Bruce R. King

DOI: https://doi.org/10.1111/dme.14511

IF: 3.5

2021-02-03

Diabetic Medicine

Abstract:AimTo determine the insulin requirement for a high fat, high protein breakfast to optimise postprandial glycaemic excursions in children and young people with type 1 diabetes using insulin pumps. MethodsTwenty‐seven participants age 10‐23 years, BMI <95th percentile (2‐18 years) or BMI <30 kg/m2 (19‐25 years) and HbA1c ≤64 mmol/mol (≤8.0%) consumed a high fat, high protein breakfast (carbohydrate:30g, fat:40g, protein:50g) on four days. In this cross‐over trial insulin was administered, based on the insulin‐to‐carbohydrate ratio (ICR) of 100% (control), 120%, 140% and 160%, in an order defined by a randomisation sequence and delivered in a combination bolus, 60% ¼ hr pre‐meal and 40% over 3 hr. Postprandial sensor glucose was assessed for 6 hr. ResultsCompared to 100%ICR, 140%ICR and 160%ICR resulted in significantly lower 6 hr areas under the glucose curves: mean (95%CI) (822 mmol/L.min [605,1039] and 567 [350,784] vs 1249 [1042,1457],p≤0.001) and peak glucose excursions (4.0 mmol/L [3.0,4.9] and 2.7 [1.7,3.6] vs 6.0 [5.0,6.9],p<0.001). Rates of hypoglycaemia for 100%‐160%ICR were 7.7%, 7.7%, 12% and 19% respectively (p≥0.139). With increasing insulin dose, a step‐wise reduction in mean glucose excursion was observed from 1‐6 hr (p=0.008). ConclusionsIncrementally increasing the insulin dose for a high fat, high protein breakfast resulted in a predictable, dose‐dependent reduction in postprandial glycaemia: 140%ICR improved postprandial glycaemic excursions without a statistically significant increase in hypoglycaemia. These findings support a safe, practical method for insulin adjustment for high fat, high protein meals that can be readily implemented in practice to improve postprandial glycaemia.

endocrinology & metabolism

[DETECTION, ON THE OCCIPITAL BONE, OF ELASTIC (ULTRASONIC) WAVES INDUCED BY LASER IMPULSES IN THE EYE OF A RABBIT].

A Randomized Clinical Trial for Meal Bolus Decision Using Learning-based Control in Adults With Type 2 Diabetes

Optimized Glycemic Control of Type 2 Diabetes with Reinforcement Learning: a Proof-of-concept Trial.

Intermittent Control for Safe Long-Acting Insulin Intensification for Type 2 Diabetes: In-Silico Experiment

Real-time Artificial Intelligence Assisted Insulin Dosage Titration System for Glucose Control in Type 2 Diabetic Patients: a Proof of Concept Study

Attention Networks for Personalized Mealtime Insulin Dosing in People with Type 1 Diabetes

Insulin dose optimization using an automated artificial intelligence-based decision support system in youths with type 1 diabetes

Basal Glucose Control in Type 1 Diabetes Using Deep Reinforcement Learning: An In Silico Validation

Managing the diversification risk.

Managing antibiotic associated diarrhoea

Use of a decision support tool and quick start onboarding tool in individuals with type 1 diabetes using advanced automated insulin delivery: a single-arm multi-phase intervention study

Reinforcement Learning-Based Adaptive Insulin Advisor for Individuals with Type 1 Diabetes Patients under Multiple Daily Injections Therapy

Simple meal announcements and pramlintide delivery versus carbohydrate counting in type 1 diabetes with automated fast-acting insulin aspart delivery: a randomised crossover trial in Montreal, Canada

Simulation-driven optimization of insulin therapy profiles in a commercial hybrid closed-loop system

Performance of an Automated Insulin Delivery System: Results of Early Phase Feasibility Studies

Reinforcement Learning: A Paradigm Shift in Personalized Blood Glucose Management for Diabetes

Real-Life Use of Automated Insulin Delivery in Individuals With Type 2 Diabetes

A Dual Mode Adaptive Basal-Bolus Advisor Based on Reinforcement Learning

Estimating a Personalized Basal Insulin Dose from Short-Term Closed-Loop Data in Type 2 Diabetes

Challenging common bolus advisor for self-monitoring type-I diabetes patients using Reinforcement Learning

In children and young people with type 1 diabetes using Pump therapy, an additional 40% of the insulin dose for a high‐fat, high‐protein breakfast improves postprandial glycaemic excursions: A cross‐over trial