Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 67 of the 67 references it cites were found in a registry or an index; 19 of the 65 sources its claims cite were checked for support.

Flexible self-rectifying synapse array for energy-efficient edge multiplication in electrocardiogram diagnosis

Open access

Published version of a preprint

Abstract

Edge computing devices, which generate, collect, process, and analyze data near the source, enhance the data processing efficiency and improve the responsiveness in real-time applications or unstable network environments. To be utilized in wearable and skin-attached electronics, these edge devices must be compact, energy efficient for use in low-power environments, and fabricable on soft substrates. Here, we propose a flexible memristive dot product engine (f-MDPE) designed for edge use and demonstrate its feasibility in a real-time electrocardiogram (ECG) monitoring system. The f-MDPE comprises a 32 × 32 crossbar array embodying a low-temperature processed self-rectifying charge trap memristor on a flexible polyimide substrate and exhibits high uniformity and robust electrical and mechanical stability even under 5-mm bending conditions. Then, we design a neural network training algorithm through hardware-aware approaches and conduct real-time edge ECG diagnosis. This approach achieved an ECG classification accuracy of 93.5%, while consuming only 0.3% of the energy compared to digital approaches, highlighting the strong potential of this approach for emerging edge neuromorphic hardware. Edge computing devices must be compact and energy efficient for use in low-power environments. Here, the authors propose a flexible memristive dot product engine designed for edge use with 0.3% energy consumption compared to digital approaches.

Overview

CoChat Summary

## TL;DR

The paper presents a flexible, self-rectifying memristive dot product engine (f-MDPE) on a polyimide substrate for edge ECG inference. It reports 93.5% ECG classification accuracy and only 0.3% of the energy of digital approaches, with stable operation under 5 mm bending.

## Key findings

  • The f-MDPE is a 32 × 32 crossbar array of low-temperature processed charge-trap memristors on flexible polyimide, and the array showed 100% yield across 160 cells (32 × 5).
  • The device demonstrated strong electrical performance, including a maximum on/off ratio of ~10^4 at 3 V, a rectifying ratio of ~10^3 at a read voltage of ±3 V, retention for 10^4 s at 85 °C, and endurance up to 10^5 cycles.
  • Under bending, the device operated well down to a 5 mm curvature radius, and retained stability over 10^3 s at r = 5 mm; performance degraded below 5 mm.
  • For ECG classification, the hardware-aware SLP on the MIT-BIH dataset achieved an F1-macro score of 0.774 and F1-micro score of 0.815 on the 4255-test set; after normal/abnormal reconstruction, the scores improved to 0.804 and 0.847, and abnormal-signal accuracy was 0.873.
  • In the real-time ECG diagnosis demonstration, the system achieved 93.5% classification accuracy while using only 0.3% of the energy of digital approaches; the estimated inference energy was 120 nJ for the 5-µm-line-width device, versus 168 µJ for CPU and 41 µJ for GPU.
  • In virtual scaling, the 2k model saturated at F1-micro 0.99 and F1-macro 0.94; the authors estimate energy could drop to 2.64 nJ with 100-nm line width.

## Methods

  • Device design/fabrication: a Pt/Ta2O5/NbOx/Al2O3/Ti/Pt/Ti stack was fabricated on a 50-µm polyimide film using low-temperature processes below 180 °C, then patterned into a 32 × 32 crossbar with 5 µm line width.
  • Data/sample: ECG classification used the MIT-BIH arrhythmia database, five classes (N, S, V, Q, F), with ECG signals resized to 32 time frames and amplitudes adjusted to 2.0–3.5 V; real-time validation used personal ECG signals from the main authors with IRB exemption and consent.
  • Analysis: the authors trained a single-layer perceptron with hardware-aware handling of non-Ohmic Schottky conduction and 16 quantized conductance states, mapped trained weights to conductance, programmed the array with ISPP verify, and tested both offline classification and a lab-scale real-time inference system; they also simulated larger virtual f-MDPE arrays with 1D CNNs.

## Limitations

  • The paper states that the prototype hardware is only 32 × 32 and that the demonstrated ECG model is a simple SLP; better performance at larger sizes was still needed, so larger arrays were simulated rather than physically fabricated.
  • The real-time ECG demonstration was explicitly not an actual diagnosis: it used a regular person and the authors say the purpose was to demonstrate system operation only.
  • The text makes evident that training on hardware was not demonstrated; the f-MDPE was programmed only for inference, and the authors note that expanding functionality beyond inference to training remains a challenge.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41467-025-59589-2

Cites

67 of 67 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Overview of a flexible memristive dot product engine (f-MDPE) embodying a self-rectifying charge-trap memristor for ECG diagnosis. a Illustration of the real-time ECG monitoring system with the f-MDPE. b Key advantages of the f-MDPE for edge AI applications.
Fig. 1 Overview of a flexible memristive dot product engine (f-MDPE) embodying a self-rectifying charge-trap memristor for ECG diagnosis. a Illustration of the real-time ECG monitoring system with the f-MDPE. b Key advantages of the f-MDPE for edge AI applications.CC BY-NC-ND · PubMed Central
Fig. 2 Electrical characteristics of the f-MDPE. a Optical images of the f-MDPE on a polyimide (PI) substrate. b Cross-sectional transmission electron microscopy (TEM) image of the device. A single representative TEM image is measured. c I–V curves of the device with various compliance currents (ICC). The inset shows the nonlinearity and on/off ratio at 10 µA of ICC. d I–V curves of 160 cells in the array under 1 µA of ICC. e Repeated potentiation and depression cycles of the device. f Retention characteristics of 8 conductance states at 85 °C. g Long-term retention characteristics at various temperatures. h Endurance performance for LRS (red) and HRS (blue) up to 105 cycles at room temperature.
Fig. 2 Electrical characteristics of the f-MDPE. a Optical images of the f-MDPE on a polyimide (PI) substrate. b Cross-sectional transmission electron microscopy (TEM) image of the device. A single representative TEM image is measured. c I–V curves of the device with various compliance currents (ICC). The inset shows the nonlinearity and on/off ratio at 10 µA of ICC. d I–V curves of 160 cells in the array under 1 µA of ICC. e Repeated potentiation and depression cycles of the device. f Retention characteristics of 8 conductance states at 85 °C. g Long-term retention characteristics at various temperatures. h Endurance performance for LRS (red) and HRS (blue) up to 105 cycles at room temperature.CC BY-NC-ND · PubMed Central
Fig. 3 Bending stability of the f-MDPE. a Memory window for various curvature radii. b Long-term retention characteristics under a 5 mm curvature radius. c Retention characteristics of 8 conductance states and d potentiation and depression under various curvature radii from flat to r = 5 mm. e Potentiation and depression after repeated bending cycles under a 5 mm radius. f–h I–V curves after repeated bending cycles under various curvature radii.
Fig. 3 Bending stability of the f-MDPE. a Memory window for various curvature radii. b Long-term retention characteristics under a 5 mm curvature radius. c Retention characteristics of 8 conductance states and d potentiation and depression under various curvature radii from flat to r = 5 mm. e Potentiation and depression after repeated bending cycles under a 5 mm radius. f–h I–V curves after repeated bending cycles under various curvature radii.CC BY-NC-ND · PubMed Central
Fig. 4 Hardware-aware training for addressing non-Ohmic conduction behavior. a Output currents of multiple states as a function of the input voltages of non-Ohmic devices following Schottky conduction (Isch). b Output currents represented by the transformed voltage, V′. g′ remains constant over the input voltages. c Crossbar array after linear transformation (V to V′ and g to g′), enabling a formal VMM operation in the f-MDPE. d Crossbar array with non-Ohmic cells. In this case, the conductance changes according to the input voltage. e Four bits (i.e., 16 levels) of multilevel conductance states at the reference voltage, 3 V, with 15 cells. Note that box plots are defined by the minimum, 25th percentile (Q1), median, 75th percentile (Q3), and maximum values, with the box indicating the interquartile range (IQR) and the whiskers representing the full data range. f I–V curves at the read voltage region at each analog state. g g′ obtained by extracting values for each analog state using a compensation constant k = 7.
Fig. 4 Hardware-aware training for addressing non-Ohmic conduction behavior. a Output currents of multiple states as a function of the input voltages of non-Ohmic devices following Schottky conduction (Isch). b Output currents represented by the transformed voltage, V′. g′ remains constant over the input voltages. c Crossbar array after linear transformation (V to V′ and g to g′), enabling a formal VMM operation in the f-MDPE. d Crossbar array with non-Ohmic cells. In this case, the conductance changes according to the input voltage. e Four bits (i.e., 16 levels) of multilevel conductance states at the reference voltage, 3 V, with 15 cells. Note that box plots are defined by the minimum, 25th percentile (Q1), median, 75th percentile (Q3), and maximum values, with the box indicating the interquartile range (IQR) and the whiskers representing the full data range. f I–V curves at the read voltage region at each analog state. g g′ obtained by extracting values for each analog state using a compensation constant k = 7.CC BY-NC-ND · PubMed Central
Fig. 5 The f-MDPE-based hardware system for ECG heartbeat diagnosis. a Schematics of in-situ inference with a pretrained f-MDPE array for ECG diagnosis. b Pretrained weight matrix of a 32 × 5 sized network, using positive and negative weights separately. c Target current at 3 V of the transferred weight matrix in (b). d Measured current at 3 V of the pretrained f-MDPE array. e Error distribution between the target and measured currents at each single point. f Summed output currents for the given input heartbeats from the MIT-BIH heartbeat dataset represented as N, S, V, Q, and F. The neuron indicating each heartbeat has the highest output current, indicating a correct classification. g, h Confusion matrices of the network for 5-category classification (g) and normal/abnormal classification (h).
Fig. 5 The f-MDPE-based hardware system for ECG heartbeat diagnosis. a Schematics of in-situ inference with a pretrained f-MDPE array for ECG diagnosis. b Pretrained weight matrix of a 32 × 5 sized network, using positive and negative weights separately. c Target current at 3 V of the transferred weight matrix in (b). d Measured current at 3 V of the pretrained f-MDPE array. e Error distribution between the target and measured currents at each single point. f Summed output currents for the given input heartbeats from the MIT-BIH heartbeat dataset represented as N, S, V, Q, and F. The neuron indicating each heartbeat has the highest output current, indicating a correct classification. g, h Confusion matrices of the network for 5-category classification (g) and normal/abnormal classification (h).CC BY-NC-ND · PubMed Central
Fig. 6 Experimental demonstration and scalability of f-MDPE for ECG diagnosis. a Photograph of the integrated system: ECG sensor, f-MDPE, and f-MDPE controller integrated on a control PC. ECG diagnosis of N pulses from the b MIT-BIH dataset and c personal signals acquired from the ECG sensor. d Schematics of the scaled-up 1D convolutional neural network (1D CNN). e Recognition rates of ECG diagnosis for the proposed 1D CNN architectures. The recognition rate is saturated in the 2k model. f Energy consumption comparisons of the f-MDPE algorithm and conventional processors (CPU and GPU) for inferring a single heartbeat signal from the 2k model. g Energy consumption per bit and area per cell comparisons of the f-MDPE and flexible or stretchable memristive arrays. The green squares represent flexible arrays, and the blue rectangles represent stretchable arrays.
Fig. 6 Experimental demonstration and scalability of f-MDPE for ECG diagnosis. a Photograph of the integrated system: ECG sensor, f-MDPE, and f-MDPE controller integrated on a control PC. ECG diagnosis of N pulses from the b MIT-BIH dataset and c personal signals acquired from the ECG sensor. d Schematics of the scaled-up 1D convolutional neural network (1D CNN). e Recognition rates of ECG diagnosis for the proposed 1D CNN architectures. The recognition rate is saturated in the 2k model. f Energy consumption comparisons of the f-MDPE algorithm and conventional processors (CPU and GPU) for inferring a single heartbeat signal from the 2k model. g Energy consumption per bit and area per cell comparisons of the f-MDPE and flexible or stretchable memristive arrays. The green squares represent flexible arrays, and the blue rectangles represent stretchable arrays.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

9 Valid · 6 Not valid · 9 Not checked · 56 Own results and interpretations

36 more claims · 3 with their cited sources checked

Create a free account to see all 80 claims and to verify the 6 references still open.

See all claims — free account
  • Probably not a substantial claim. 16
  • It may state two separate claims as one. 27
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Citations

Cites

Show all 67 references

Cited by

See all 19

Metadata

Authors
Lee, Y., Rhee, H., Kim, G., Cheong, W. H., Kim, D. H., Song, H., Kay, S. N., Lee, J., & Kim, K. M.
Affiliations
  • Korea Advanced Institute of Science and Technology (KR)
  • Chungnam National University (KR)
Corresponding author
  • Lee, Jongwon
  • Kim, Kyung Min
Venue
Nature Communications
Published
2025-05-09
Volume
16
Issue
1
Pages
4312-4312
ISSN
2041-1723 (ISSN-L)
OpenAlex
W4410197042
Cited by
19
References
67
Citations by year
2026: 10 · 2025: 9
FWCI
4.22
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by-nc-nd)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • National Research Foundation of Korea (RS-2023-00216619, 2022M3F3A2A01076569, 2022M3I7A4085484, 2023R1A2C2005159)
  • National Nano Fab Center (NNFC) : National Nano Fab Center (NNFC) is affiliated institution of KAIST.
Topics
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Analog and Mixed-Signal Circuit Design
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Ferroelectric and Negative Capacitance Devices
Keywords
flexible memristors, crossbar arrays, edge computing, neuromorphic hardware, ECG classification, ECG monitoring, energy-efficient hardware, polyimide substrate, wearable electronics, soft electronics, energy-efficient computing
MeSH terms
Wearable Electronic Devices, Algorithms, Electrocardiography, Equipment Design, Humans, Signal Processing, Computer-Assisted, Neural Networks, Computer

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Flexible self-rectifying synapse array for energy-efficient edge multiplication in electrocardiogram diagnosis. https://app.cochat.ai/papers/10.1038/s41467-025-59589-2
The paper
Lee, Y., Rhee, H., Kim, G., Cheong, W. H., Kim, D. H., Song, H., Kay, S. N., Lee, J., & Kim, K. M. (2025). Flexible self-rectifying synapse array for energy-efficient edge multiplication in electrocardiogram diagnosis. Nature Communications. https://doi.org/10.1038/s41467-025-59589-2

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.