arXiv subjects

Kezhi Li

Publications and source records attributed to Kezhi Li.

35 records · Page 2Linked to original sources

arXiv2024

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

Newly diagnosed Type 1 Diabetes (T1D) patients often struggle to obtain effective Blood Glucose (BG) prediction models due to the lack of sufficient BG data from Continuous Glucose Monitoring (CGM), presenting a significant "cold start" problem in patient care. Utilizing population models to address this challenge is a potential solution, but collecting patient data for training population models in a privacy-conscious manner is challenging, especially given that such data is often stored on personal devices. Considering the privacy protection and addressing the "cold start" problem in diabetes care, we propose "GluADFL", blood Glucose prediction by Asynchronous Decentralized Federated Learning. We compared GluADFL with eight baseline methods using four distinct T1D datasets, comprising 298 participants, which demonstrated its superior performance in accurately predicting BG levels for cross-patient analysis. Furthermore, patients' data might be stored and shared across various communication networks in GluADFL, ranging from highly interconnected (e.g., random, performs the best among others) to more structured topologies (e.g., cluster and ring), suitable for various social networks. The asynchronous training framework supports flexible participation. By adjusting the ratios of inactive participants, we found it remains stable if less than 70% are inactive. Our results confirm that GluADFL offers a practical, privacy-preserving solution for BG prediction in T1D, significantly enhancing the quality of diabetes management.

cs.LG↗

arXiv2024

GARNN: An Interpretable Graph Attentive Recurrent Neural Network for Predicting Blood Glucose Levels via Multivariate Time Series

Accurate prediction of future blood glucose (BG) levels can effectively improve BG management for people living with diabetes, thereby reducing complications and improving quality of life. The state of the art of BG prediction has been achieved by leveraging advanced deep learning methods to model multi-modal data, i.e., sensor data and self-reported event data, organised as multi-variate time series (MTS). However, these methods are mostly regarded as ``black boxes'' and not entirely trusted by clinicians and patients. In this paper, we propose interpretable graph attentive recurrent neural networks (GARNNs) to model MTS, explaining variable contributions via summarizing variable importance and generating feature maps by graph attention mechanisms instead of post-hoc analysis. We evaluate GARNNs on four datasets, representing diverse clinical scenarios. Upon comparison with twelve well-established baseline methods, GARNNs not only achieve the best prediction accuracy but also provide high-quality temporal interpretability, in particular for postprandial glucose levels as a result of corresponding meal intake and insulin injection. These findings underline the potential of GARNN as a robust tool for improving diabetes care, bridging the gap between deep learning technology and real-world healthcare solutions.

cs.LG↗

arXiv2023

Going faster to see further: GPU-accelerated value iteration and simulation for perishable inventory control using JAX

Value iteration can find the optimal replenishment policy for a perishable inventory problem, but is computationally demanding due to the large state spaces that are required to represent the age profile of stock. The parallel processing capabilities of modern GPUs can reduce the wall time required to run value iteration by updating many states simultaneously. The adoption of GPU-accelerated approaches has been limited in operational research relative to other fields like machine learning, in which new software frameworks have made GPU programming widely accessible. We used the Python library JAX to implement value iteration and simulators of the underlying Markov decision processes in a high-level API, and relied on this library's function transformations and compiler to efficiently utilize GPU hardware. Our method can extend use of value iteration to settings that were previously considered infeasible or impractical. We demonstrate this on example scenarios from three recent studies which include problems with over 16 million states and additional problem features, such as substitution between products, that increase computational complexity. We compare the performance of the optimal replenishment policies to heuristic policies, fitted using simulation optimization in JAX which allowed the parallel evaluation of multiple candidate policy parameters on thousands of simulated years. The heuristic policies gave a maximum optimality gap of 2.49%. Our general approach may be applicable to a wide range of problems in operational research that would benefit from large-scale parallel computation on consumer-grade GPU hardware.

cs.AI↗

arXiv2020

A novel hand-held interface supporting the self-management of Type 1 diabetes

The paper describes the interaction design of a hand-held interface supporting the self-management of Type 1 diabetes. It addresses well-established clinical and human-computer interaction requirements. The design exploits three opportunities. One is associated with visible context, whether conspicuous or inconspicuous. A second arises from the design freedom made possible by the user's anticipated focus of attention during certain interactions. A third opportunity to provide valuable functionality arises from wearable sensors and machine learning algorithms. The resulting interface permits ``What if?'' questions: it allows a user to dynamically and manually explore predicted short-term (e.g., 2 hours) relationships between an intended meal, blood glucose level and recommended insulin dosage, and thereby readily make informed food and exercise decisions. Design activity has been informed throughout by focus groups comprising people with Type 1 diabetes in addition to experts in diabetes, interaction design and machine learning. The design is being implemented prior to a clinical trial.

cs.CY↗

arXiv2020

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

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 65.9% 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.

eess.SP↗

arXiv2019

A Dual-Hormone Closed-Loop Delivery System for Type 1 Diabetes Using Deep Reinforcement Learning

We propose a dual-hormone delivery strategy by exploiting deep reinforcement learning (RL) for people with Type 1 Diabetes (T1D). Specifically, double dilated recurrent neural networks (RNN) are used to learn the hormone delivery strategy, trained by a variant of Q-learning, whose inputs are raw data of glucose \& meal carbohydrate and outputs are dual-hormone (insulin and glucagon) delivery. Without prior knowledge of the glucose-insulin metabolism, we run the method on the UVA/Padova simulator. Hundreds days of self-play are performed to obtain a generalized model, then importance sampling is adopted to customize the model for personal use. \emph{In-silico} the proposed strategy achieves glucose time in target range (TIR) $93\%$ for adults and $83\%$ for adolescents given standard bolus, outperforming previous approaches significantly. The results indicate that deep RL is effective in building personalized hormone delivery strategy for people with T1D.

q-bio.QM↗

arXiv2019

Convolutional Recurrent Neural Networks for Glucose Prediction

Control of blood glucose is essential for diabetes management. Current digital therapeutic approaches for subjects with Type 1 diabetes mellitus (T1DM) such as the artificial pancreas and insulin bolus calculators leverage machine learning techniques for predicting subcutaneous glucose for improved control. Deep learning has recently been applied in healthcare and medical research to achieve state-of-the-art results in a range of tasks including disease diagnosis, and patient state prediction among others. In this work, we present a deep learning model that is capable of forecasting glucose levels with leading accuracy for simulated patient cases (RMSE = 9.38$\pm$0.71 [mg/dL] over a 30-minute horizon, RMSE = 18.87$\pm$2.25 [mg/dL] over a 60-minute horizon) and real patient cases (RMSE = 21.07$\pm$2.35 [mg/dL] for 30-minute, RMSE = 33.27$\pm$4.79\% for 60-minute). In addition, the model provides competitive performance in providing effective prediction horizon ($PH_{eff}$) with minimal time lag both in a simulated patient dataset ($PH_{eff}$ = 29.0$\pm$0.7 for 30-min and $PH_{eff}$ = 49.8$\pm$2.9 for 60-min) and in a real patient dataset ($PH_{eff}$ = 19.3$\pm$3.1 for 30-min and $PH_{eff}$ = 29.3$\pm$9.4 for 60-min). This approach is evaluated on a dataset of 10 simulated cases generated from the UVa/Padova simulator and a clinical dataset of 10 real cases each containing glucose readings, insulin bolus, and meal (carbohydrate) data. Performance of the recurrent convolutional neural network is benchmarked against four algorithms. The proposed algorithm is implemented on an Android mobile phone, with an execution time of $6$ms on a phone compared to an execution time of $780$ms on a laptop.

cs.CV↗

arXiv2018

Two-qubit state recovery from amplitude damping based on weak measurement

In the quantum control process, arbitrary pure or mixed initial states need to be protected from amplitude damping through the noise channel using measurements and quantum control. However, how to achieve it on a two-qubit quantum system remains a challenge. In this paper, we propose a feed-forward control approach to protect arbitrary two-qubit pure or mixed initial states using the weak measurement. A feed-forward operation and measurements are used before the noise channel, and afterwards a reversed operation and measurements are applied to recover the state back to its initial state. In the case of two-qubit pure states, we use the unravelling trick to describe the state of the system in each step of the control procedure. For two-qubit mixed states, a completely-positive trace-preserving (CPTP) map is implemented. Finally, the fidelity and success probability are used to evaluate the effect of protection. The complete recovery conditions for the measurement strengths are derived, under which we achieve the optimal fidelity and the success probability of recovering the initial pure or mixed states.

quant-ph↗

arXiv2016

A Nonlinear Weighted Total Variation Image Reconstruction Algorithm for Electrical Capacitance Tomography

A new iterative image reconstruction algorithm for electrical capacitance tomography (ECT) is proposed that is based on iterative soft thresholding of a total variation penalty and adaptive reweighted compressive sensing. This algorithm encourages sharp changes in the ECT image and overcomes the disadvantage of the $l_1$ minimization by equipping the total variation with an adaptive weighting depending on the reconstructed image. Moreover, the non-linear effect is also partially reduced due to the adoption of an updated sensitivity matrix. Simulation results show that the proposed algorithm recovers ECT images more precisely than existing state-of-the-art algorithms and therefore is suitable for the imaging of multiphase systems in industrial or medical applications.

cs.CV↗

arXiv2016

A balanced rail-to-rail all digital comparator using only standard cells

An all-digital comparator with full input range is presented. It outperforms the nowaday all-digital comparators with its large rail-to-rail input range. This is achieved by the proposed Yin-yang balance mechanism between the two logic gates: NAND3 and OAI (Or-And-Invert). The important design considerations to achieve this balance are presented, such as the driving strength manipulation and the use of pre-distortion technique. Constructed only by commercially available digital standard cells, the layout of the proposed comparator is generated automatically by standard digital Place & Route routine within several minutes. The Verilog code for the proposed circuit is given, and the circuit is successfully implemented in 130nm CMOS technology with the power consumption of 0.176mW at the clock of 330MHz.

cs.OH↗

arXiv2015

A Brief Survey of Image Processing Algorithms in Electrical Capacitance Tomography

To study the fundamental physics of complex multiphase flow systems using advanced measurement techniques, especially the electrical capacitance tomography (ECT) approach, this article carries out an initial literature review of the ECT method from a point of view of signal processing and algorithm design. After introducing the physical laws governing the ECT system, we will focus on various reconstruction techniques that are capable to recover the image of the internal characteristics of a specified region based on the measuring capacitances of multi-electrode sensors surrounding the region. Each technique has its own advantages and limitations, and many algorithms have been examined by simulations or experiments. Future researches in 3D reconstruction and other potential improvements of the system are discussed in the end.

cs.CV↗

arXiv2014

State of the Art and Prospects of Structured Sensing Matrices in Compressed Sensing

Compressed sensing (CS) enables people to acquire the compressed measurements directly and recover sparse or compressible signals faithfully even when the sampling rate is much lower than the Nyquist rate. However, the pure random sensing matrices usually require huge memory for storage and high computational cost for signal reconstruction. Many structured sensing matrices have been proposed recently to simplify the sensing scheme and the hardware implementation in practice. Based on the restricted isometry property and coherence, couples of existing structured sensing matrices are reviewed in this paper, which have special structures, high recovery performance, and many advantages such as the simple construction, fast calculation and easy hardware implementation. The number of measurements and the universality of different structure matrices are compared.

cs.IT↗

arXiv2014

Alternating Strategies Are Good For Low-Rank Matrix Reconstruction

This article focuses on the problem of reconstructing low-rank matrices from underdetermined measurements using alternating optimization strategies. We endeavour to combine an alternating least-squares based estimation strategy with ideas from the alternating direction method of multipliers (ADMM) to recover structured low-rank matrices, such as Hankel structure. We show that merging these two alternating strategies leads to a better performance than the existing alternating least squares (ALS) strategy. The performance is evaluated via numerical simulations.

math.ST↗

arXiv2014

Piecewise Toeplitz Matrices-based Sensing for Rank Minimization

This paper proposes a set of piecewise Toeplitz matrices as the linear mapping/sensing operator $\mathcal{A}: \mathbf{R}^{n_1 \times n_2} \rightarrow \mathbf{R}^M$ for recovering low rank matrices from few measurements. We prove that such operators efficiently encode the information so there exists a unique reconstruction matrix under mild assumptions. This work provides a significant extension of the compressed sensing and rank minimization theory, and it achieves a tradeoff between reducing the memory required for storing the sampling operator from $\mathcal{O}(n_1n_2M)$ to $\mathcal{O}(\max(n_1,n_2)M)$ but at the expense of increasing the number of measurements by $r$. Simulation results show that the proposed operator can recover low rank matrices efficiently with a reconstruction performance close to the cases of using random unstructured operators.

cs.IT↗

arXiv2014

A Robust Compressive Quantum State Tomography Algorithm Using ADMM

The possible state space dimension increases exponentially with respect to the number of qubits. This feature makes the quantum state tomography expensive and impractical for identifying the state of merely several qubits. The recent developed approach, compressed sensing, gives us an alternative to estimate the quantum state with fewer measurements. It is proved that the estimation then can be converted to a convex optimization problem with quantum mechanics constraints. In this paper we present an alternating augmented Lagrangian method for quantum convex optimization problem aiming for recovering pure or near pure quantum states corrupted by sparse noise given observables and the expectation values of the measurements. The proposed algorithm is much faster, robust to outlier noises (even very large for some entries) and can solve the reconstruction problem distributively. The simulations verify the superiority of the proposed algorithm and compare it to the conventional least square and compressive quantum tomography using Dantzig method.

cs.IT↗

arXiv2012

Convolutional Compressed Sensing Using Deterministic Sequences

In this paper, a new class of circulant matrices built from deterministic sequences is proposed for convolution-based compressed sensing (CS). In contrast to random convolution, the coefficients of the underlying filter are given by the discrete Fourier transform of a deterministic sequence with good autocorrelation. Both uniform recovery and non-uniform recovery of sparse signals are investigated, based on the coherence parameter of the proposed sensing matrices. Many examples of the sequences are investigated, particularly the Frank-Zadoff-Chu (FZC) sequence, the \textit{m}-sequence and the Golay sequence. A salient feature of the proposed sensing matrices is that they can not only handle sparse signals in the time domain, but also those in the frequency and/or or discrete-cosine transform (DCT) domain.

cs.IT↗

arXiv2012

Orthogonal symmetric Toeplitz matrices for compressed sensing: Statistical isometry property

Recently, the statistical restricted isometry property (RIP) has been formulated to analyze the performance of deterministic sampling matrices for compressed sensing. In this paper, we propose the usage of orthogonal symmetric Toeplitz matrices (OSTM) for compressed sensing and study their statistical RIP by taking advantage of Stein's method. In particular, we derive the statistical RIP performance bound in terms of the largest value of the sampling matrix and the sparsity level of the input signal. Based on such connections, we show that OSTM can satisfy the statistical RIP for an overwhelming majority of signals with given sparsity level, if a Golay sequence used to generate the OSTM. Such sensing matrices are deterministic, Toeplitz, and efficient to implement. Simulation results show that OSTM can offer reconstruction performance similar to that of random matrices.

cs.IT↗