RESUMEN
In the past 50 years, the advent of electronic technology to directly interface with neural tissue has transformed the fields of medicine and biology. Devices that restore or even replace impaired bodily functions, such as deep brain stimulators and cochlear implants, have ushered in a new treatment era for previously intractable conditions. Meanwhile, electrodes for recording and stimulating neural activity have allowed researchers to unravel the vast complexities of the human nervous system. Recent advances in semiconducting materials have allowed effective interfaces between electrodes and neuronal tissue through novel devices and structures. Often these are unattainable using conventional metallic electrodes. These have translated into advances in research and treatment. The development of semiconducting materials opens new avenues in neural interfacing. This review considers this emerging class of electrodes and how it can facilitate electrical, optical, and chemical sensing and modulation with high spatial and temporal precision. Semiconducting electrodes have advanced electrically based neural interfacing technologies owing to their unique electrochemical and photo-electrochemical attributes. Key operation modalities, namely sensing and stimulation in electrical, biochemical, and optical domains, are discussed, highlighting their contrast to metallic electrodes from the application and characterization perspective.
Asunto(s)
Sistema Nervioso , Neuronas , Humanos , Electrodos , Neuronas/fisiología , ElectricidadRESUMEN
There has been a growing interest in computational electroencephalogram (EEG) signal processing in a diverse set of domains, such as cortical excitability analysis, event-related synchronization, or desynchronization analysis. In recent years, several inconsistencies were found across different EEG studies, which authors often attributed to methodological differences. However, the assessment of such discrepancies is deeply underexplored. It is currently unknown if methodological differences can fully explain emerging differences and the nature of these differences. This study aims to contrast widely used methodological approaches in EEG processing and compare their effects on the outcome variables. To this end, two publicly available datasets were collected, each having unique traits so as to validate the results in two different EEG territories. The first dataset included signals with event-related potentials (visual stimulation) from 45 subjects. The second dataset included resting state EEG signals from 16 subjects. Five EEG processing steps, involved in the computation of power and phase quantities of EEG frequency bands, were explored in this study: artifact removal choices (with and without artifact removal), EEG signal transformation choices (raw EEG channels, Hjorth transformed channels, and averaged channels across primary motor cortex), filtering algorithms (Butterworth filter and Blackman-Harris window), EEG time window choices (-750 ms to 0 ms and -250 ms to 0 ms), and power spectral density (PSD) estimation algorithms (Welch's method, Fast Fourier Transform, and Burg's method). Powers and phases estimated by carrying out variations of these five methods were analyzed statistically for all subjects. The results indicated that the choices in EEG transformation and time-window can strongly affect the PSD quantities in a variety of ways. Additionally, EEG transformation and filter choices can influence phase quantities significantly. These results raise the need for a consistent and standard EEG processing pipeline for computational EEG studies. Consistency of signal processing methods cannot only help produce comparable results and reproducible research, but also pave the way for federated machine learning methods, e.g., where model parameters rather than data are shared.
Asunto(s)
Electroencefalografía , Procesamiento de Señales Asistido por Computador , Algoritmos , Potenciales Evocados , Análisis de Fourier , HumanosRESUMEN
Donor doping of perovskite oxides has emerged as an attractive technique to create high performance and low energy non-volatile analog memories. Here, we examine the origins of improved switching performance and stable multi-state resistive switching in Nb-doped oxygen-deficient amorphous SrTiO3 (Nb:a-STO x ) metal-insulator-metal (MIM) devices. We probe the impact of substitutional dopants (i.e., Nb) in modulating the electronic structure and subsequent switching performance. Temperature stability and bias/time dependence of the switching behavior are used to ascertain the role of substitutional dopants and highlight their utility to modulate volatile and non-volatile behavior in a-STO x devices for adaptive and neuromorphic applications. We utilized a combination of transmission electron microscopy, photoluminescence emission properties, interfacial compositional evaluation, and activation energy measurements to investigate the microstructure of the nanofilamentary network responsible for switching. These results provide important insights into understanding mechanisms that govern the performance of donor-doped perovskite oxide-based memristive devices.
RESUMEN
PURPOSE: This study introduces an algorithm specifically designed for processing unprocessed 12-lead electrocardiogram (ECG) data, with the primary aim of detecting cardiac abnormalities. METHODS: The proposed model integrates Diagonal State Space Sequence (S4D) model into its architecture, leveraging its effectiveness in capturing dynamics within time-series data. The S4D model is designed with stacked S4D layers for processing raw input data and a simplified decoder using a dense layer for predicting abnormality types. Experimental optimization determines the optimal number of S4D layers, striking a balance between computational efficiency and predictive performance. This comprehensive approach ensures the model's suitability for real-time processing on hardware devices with limited capabilities, offering a streamlined yet effective solution for heart monitoring. RESULTS: Among the notable features of this algorithm is its strong resilience to noise, enabling the algorithm to achieve an average F1-score of 81.2% and an AUROC of 95.5% in generalization. The model underwent testing specifically on the lead II ECG signal, exhibiting consistent performance with an F1-score of 79.5% and an AUROC of 95.7%. CONCLUSION: It is characterized by the elimination of pre-processing features and the availability of a low-complexity architecture that makes it suitable for implementation on numerous computing devices because it is easily implementable. Consequently, this algorithm exhibits considerable potential for practical applications in analyzing real-world ECG data. This model can be placed on the cloud for diagnosis. The model was also tested on lead II of the ECG alone and has demonstrated promising results, supporting its potential for on-device application.
Asunto(s)
Algoritmos , Electrocardiografía , Valor Predictivo de las Pruebas , Procesamiento de Señales Asistido por Computador , Humanos , Frecuencia Cardíaca , Reproducibilidad de los Resultados , Factores de Tiempo , Modelos Cardiovasculares , Arritmias Cardíacas/fisiopatología , Arritmias Cardíacas/diagnóstico , Arritmias Cardíacas/clasificación , Potenciales de Acción , Diagnóstico por ComputadorRESUMEN
Chemiresistive polymer-based sensors are promising platforms for monitoring various gases and volatile organic compounds. While they offer appealing attributes, such as ease of fabrication, flexibility, and cost-effectiveness, most of these sensors have a nearly identical response to cross-reactive gases, such as ammonia (NH3) and carbon dioxide (CO2). Aiming to address the shortcomings of chemiresistive polymer-based sensors in selectivity and simultaneous measurements of cross-reactive gases, a chemiresistive sensor array was developed consisting of components sensitive to carbon dioxide and ammonia as well as a control segment to provide the baseline. The designed system demonstrated a wide detection range for both ammonia (ranging from 0.05 to 1000 ppm) and carbon dioxide (ranging from 103 to 106 ppm) at both room and low temperatures (e.g., 4 °C). Our results also demonstrate the ability of this sensor array for the simultaneous detection of carbon dioxide and ammonia selectively in the presence of other gases and volatile organic compounds. Finally, the array was used to monitor CO2/NH3 in real food samples to demonstrate the potential for real-world applications.
Asunto(s)
Amoníaco , Dióxido de Carbono , Amoníaco/análisis , Dióxido de Carbono/análisis , Gases/análisis , Gases/químicaRESUMEN
Aim: Deep learning's widespread use prompts heightened scrutiny, particularly in the biomedical fields, with a specific focus on model generalizability. This study delves into the influence of training data characteristics on the generalization performance of models, specifically in cardiac abnormality detection. Materials & methods: Leveraging diverse electrocardiogram datasets, models are trained on subsets with varying characteristics and subsequently compared for performance. Additionally, the introduction of the attention mechanism aims to improve generalizability. Results: Experiments reveal that using a balanced dataset, just 1% of a large dataset, leads to equal performance in generalization tasks, notably in detecting cardiology abnormalities. Conclusion: This balanced training data notably enhances model generalizability, while the integration of the attention mechanism further refines the model's ability to generalize effectively.
This study tackles a common problem for deep learning models: they often struggle when faced with new, unfamiliar data that they have not been trained on. This phenomenon is also known as performance drop in out-of-distribution generalization. This reduced performance on out-of-distribution generalization is a key focus of the research, aiming to improve the models' ability to handle diverse data sets beyond their training data.The study examines how the characteristics of the dataset used to train deep learning models affect their ability to detect abnormal heart activities when applied to new, unseen data. Researchers trained these models using various sets of electrocardiogram (ECG) data and then evaluated their performance in identifying abnormalities. They also introduced an attention mechanism to enhance the models' learning capabilities. The attention mechanism in deep learning is like a spotlight that helps the model focus on important information while ignoring less relevant details.The findings were particularly noteworthy. Despite being trained on a small, well-balanced subset of a larger dataset, the models excelled in detecting heart abnormalities in new, unfamiliar data. This training method significantly improved the models' generalization and performance with unseen data. Furthermore, integrating the attention mechanism substantially enhanced the models' ability to generalize effectively on new information.
Asunto(s)
Aprendizaje Profundo , Electrocardiografía , Humanos , Electrocardiografía/métodosRESUMEN
This study introduces two models, ConvLSTM2D-liquid time-constant network (CLTC) and ConvLSTM2D-closed-form continuous-time neural network (CCfC), designed for abnormality identification using electrocardiogram (ECG) data. Trained on the Telehealth Network of Minas Gerais (TNMG) subset dataset, both models were evaluated for their performance, generalizability capacity, and resilience. They demonstrated comparable results in terms of F1 scores and AUROC values. The CCfC model achieved slightly higher accuracy, while the CLTC model showed better handling of empty channels. Remarkably, the models were successfully deployed on a resource-constrained microcontroller, proving their suitability for edge device applications. Generalization capabilities were confirmed through the evaluation on the China Physiological Signal Challenge 2018 (CPSC) dataset. The models' efficient resource utilization, occupying 70.6% of memory and 9.4% of flash memory, makes them promising candidates for real-world healthcare applications. Overall, this research advances abnormality identification in ECG data, contributing to the progress of AI in healthcare.
Asunto(s)
Electrocardiografía , Valor Predictivo de las Pruebas , Procesamiento de Señales Asistido por Computador , Humanos , Electrocardiografía/instrumentación , Redes Neurales de la Computación , Reproducibilidad de los Resultados , Frecuencia Cardíaca , Bases de Datos Factuales , Diagnóstico por Computador/instrumentación , Arritmias Cardíacas/diagnóstico , Arritmias Cardíacas/fisiopatología , Factores de Tiempo , Diseño de Equipo , Potenciales de AcciónRESUMEN
This paper proposes a programmable inhibitory interconnection network between pixels in an array of novel low-voltage Schmitt-trigger-based PFM sensors that will be of interest for future applications in memristor-based early vision processing. In addition, a new low-power inverter-based pulse-frequency modulation (PFM) design and its integration with the network is also presented. To ensure no change in the memristors conductance in the network, the CMOS imager was designed for low voltage operation. That has resulted in a significant power reduction, better than 60%, and a comparable linear dynamic range when compared to published designs in the literature. The design was performed using a 0.13 um Samsung Electronics standard CMOS process, using 0.75 V supply voltage.
Asunto(s)
Equipos de Almacenamiento de Computador , Nanotecnología/instrumentación , Fotograbar/instrumentación , Procesamiento de Señales Asistido por Computador/instrumentación , Transductores , Impedancia Eléctrica , Diseño de Equipo , Análisis de Falla de EquipoRESUMEN
This paper introduces an integrated sensor circuit based on an analog Memristor-MOS (M2) pattern matching building block that calculates the similarity/dissimilarity between two analog values. A new approach for a pulse-width modulation pixel image sensor compatible with the memristive-MOS matching structure is introduced allowing direct comparison between incoming and stored images. The pulsed-width encoded information from the pixels is forwarded to a matching circuitry that provides an anti-Gaussian-like comparison between the states of memristors. The non-volatile and multi-state memory characteristics of memristor, together with the related ability to be programmed at any one of the intermediate states between logic '1' and logic '0' brings us closer to the implementation of bio-machines that can eventually emulate human-like sensory functions.
Asunto(s)
Biomimética/instrumentación , Equipos de Almacenamiento de Computador , Interpretación de Imagen Asistida por Computador/instrumentación , Reconocimiento de Normas Patrones Automatizadas/métodos , Procesamiento de Señales Asistido por Computador/instrumentación , Impedancia Eléctrica , Diseño de Equipo , Análisis de Falla de Equipo , Humanos , Integración de SistemasRESUMEN
Objective.This study presents a proof-of-concept optical telemetry module that leverages a single light-emitting diode (LED) to transmit data at a high bit rate while consuming low power and occupying a small area. Our experiments showed that we could achieve 108 Mbit s-1and 54 Mbit s-1back telemetry data rates for tissue thicknesses of 3 mm and 8 mm, respectively.Approach.The proposed module is designed to be powered by near-field coupling and achieve bidirectional communication by low-speed downlink from near-field communication. It aims to minimize the size of the implant while providing reliable transmission that meets the requirements of high-speed wireless communication from a multi-electrode array neurotechnology implant outside the body.Results.The power consumption of the module is 1.57 mW, including the power consumption of related circuits, resulting in an efficiency of 14.5 pJ bit-1, at a tissue thickness of 3 mm and a data rate of 108 Mbit. The use of an optical lens, combined with tissue scattering effect and optimized emission angle, makes the module robust to misalignments of up to ±5 mm and ±15° between the implantable and external units. The LED in the implantable unit is only 0.98 × 0.98 × 0.6 mm3, and the testing module is composed of discrete components and laboratory instruments.Significance.This work aims to show how it is possible to strike a balance between a small, reliable, and high-bit-rate data uplink between a neural implant and its proximal, wirelessly connected external unit. This optical telemetry module has the potential to be integrated into a significantly miniaturized system through an application-specific integrated circuit and can support up to 1000 channels of neural recordings, each sampled at 9 kSps with a 12-bit readout resolution.
Asunto(s)
Amplificadores Electrónicos , Telemetría , Diseño de Equipo , Electrodos Implantados , Tecnología InalámbricaRESUMEN
Epilepsy is a prevalent condition characterized by recurrent, unpredictable seizures. Monitoring with surface electroencephalography (EEG) is the gold standard for diagnosing epilepsy, but a time-consuming, uncomfortable and sometimes ineffective process for patients. Further, using EEG over a brief monitoring period has variable success, dependent on patient tolerance and seizure frequency. The availability of hospital resources and hardware and software specifications inherently restrict the options for comfortable, long-term data collection, resulting in limited data for training machine-learning models. This mini-review examines the current patient journey, providing an overview of the current state of EEG monitoring with reduced electrodes and automated channel reduction methods. Opportunities for improving data reliability through multi-modal data fusion are suggested. We assert the need for further research in electrode reduction to advance brain monitoring solutions towards portable, reliable devices that simultaneously offer patient comfort, perform ultra-long-term monitoring and expedite the diagnosis process.
RESUMEN
This paper studies the possibility of heart kinetic motion for designing a self-powered intracardiac leadless pacemaker by piezoelectric energy harvesting. A Doppler laser displacement sensor measures in vivo heart kinetic motion. Cantilevered and four-point bending piezoelectric harvesters are studied under the measured in vivo heart kinetic motion. The heart movement is above 15 mm. The cantilevered and four-point bending harvesters generate a maximum voltage of ~ 0.28 V and 0.8 V, respectively with the measured heart motion with a heart rate of 168 beats per minute. Two DC/DC converters, LTC3588 and MAX17220, combined with full-bridge rectifiers and their start-up performance are tested.Clinical Relevance-This paper analyzed the heart kinetic motion and establishes the piezoelectric energy harvesting for a new era of self-powered leadless pacemakers.
Asunto(s)
Marcapaso Artificial , Corazón , Prótesis e Implantes , Suministros de Energía Eléctrica , TecnologíaRESUMEN
The application of deep learning models to evaluate connectome data is gaining interest in epilepsy research. Deep learning may be a useful initial tool to partition connectome data into network subsets for further analysis. Few prior works have used deep learning to examine structural connectomes from patients with focal epilepsy. We evaluated whether a deep learning model applied to whole-brain connectomes could classify 28 participants with focal epilepsy from 20 controls and identify nodal importance for each group. Participants with epilepsy were further grouped based on whether they had focal seizures that evolved into bilateral tonic-clonic seizures (17 with, 11 without). The trained neural network classified patients from controls with an accuracy of 72.92%, while the seizure subtype groups achieved a classification accuracy of 67.86%. In the patient subgroups, the nodes and edges deemed important for accurate classification were also clinically relevant, indicating the model's interpretability. The current work expands the evidence for the potential of deep learning to extract relevant markers from clinical datasets. Our findings offer a rationale for further research interrogating structural connectomes to obtain features that can be biomarkers and aid the diagnosis of seizure subtypes.
RESUMEN
Biopotential signals, like electrocardiography (ECG), electromyography (EMG), and electroencephalography (EEG), can help diagnose cardiological, musculoskeletal and neurological disorders. Dry silver/silver chloride (Ag/AgCl) electrodes are commonly used to obtain these signals. While a conductive hydrogel can be added to Ag/AgCl electrodes to improve the contact and adhesion between the electrode and the skin, dry electrodes are prone to movement. Considering that the conductive hydrogel dries over time, the use of these electrodes often creates an imbalanced skin-electrode impedance and a number of sensing issues in the front-end analogue circuit. This issue can be extended to several other electrode types that are commonly in use, in particular, for applications with a need for long-term wearable monitoring such as ambulatory epilepsy monitoring. Liquid metal alloys, such as eutectic gallium indium (EGaIn), can address key critical requirements around consistency and reliability but present challenges on low viscosity and the risk of leakage. To solve these problems, here, we demonstrate the use of a non-eutectic Ga-In alloy as a shear-thinning non-Newtonian fluid to offer superior performance to commercial hydrogel electrodes, dry electrodes, and conventional liquid metals for electrography measurements. This material has high viscosity when still and can flow like a liquid metal when sheared, preventing leakage while allowing the effective fabrication of electrodes. Moreover, the Ga-In alloy not only has good biocompatibility but also offers an outstanding skin-electrode interface, allowing for the long-term acquisition of high-quality biosignals. The presented Ga-In alloy is a superior alternative to conventional electrode materials for real-world electrography or bioimpedance measurement.
Asunto(s)
Técnicas Biosensibles , Reproducibilidad de los Resultados , Electrodos , Impedancia Eléctrica , Aleaciones , Indio , Electrocardiografía , HidrogelesRESUMEN
For the care of neonatal infants, abdominal auscultation is considered a safe, convenient, and inexpensive method to monitor bowel conditions. With the help of early automated detection of bowel dysfunction, neonatologists could create a diagnosis plan for early intervention. In this article, a novel technique is proposed for automated peristalsis sound detection from neonatal abdominal sound recordings and compared to various other machine learning approaches. It adopts an ensemble approach that utilises handcrafted as well as one and two dimensional deep features obtained from Mel Frequency Cepstral Coefficients (MFCCs). The results are then refined with the help of a hierarchical Hidden Semi-Markov Models (HSMM) strategy. We evaluate our method on abdominal sounds collected from 49 newborn infants admitted to our tertiary Neonatal Intensive Care Unit (NICU). The results of leave-one-patient-out cross validation show that our method provides an accuracy of 95.1% and an Area Under Curve (AUC) of 85.6%, outperforming both the baselines and the recent works significantly. These encouraging results show that our proposed Ensemble-based Deep Learning model is helpful for neonatologists to facilitate tele-health applications.
Asunto(s)
Auscultación , Aprendizaje Automático , Recién Nacido , Lactante , Humanos , Unidades de Cuidado Intensivo NeonatalRESUMEN
This paper proposes an artificial intelligence system that continuously improves over time at event prediction using initially unlabelled data by using self-supervised learning. Time-series data are inherently autocorrelated. By using a detection model to generate weak labels on the fly, which are concurrently used as targets to train a prediction model on a time-shifted input data stream, this autocorrelation can effectively be harnessed to reduce the burden of manual labelling. This is critical in medical patient monitoring, as it enables the development of personalized forecasting models without demanding the annotation of long sequences of physiological signal recordings. We perform a feasibility study on seizure prediction, which is identified as an ideal test case, as pre-ictal brainwaves are patient-specific, and tailoring models to individual patients is known to improve forecasting performance significantly. Our self-supervised approach is used to train individualized forecasting models for 10 patients, showing an average relative improvement in sensitivity by 14.30% and a reduction in false alarms by 19.61% in early seizure forecasting. This proof-of-concept on the feasibility of using a continuous stream of time-series neurophysiological data paves the way towards a low-power neuromorphic neuromodulation system.
RESUMEN
Technological advancements in high-throughput genomics enable the generation of complex and large data sets that can be used for classification, clustering, and bio-marker identification. Modern deep learning algorithms provide us with the opportunity of finding most significant features in such huge dataset to characterize diseases (e.g., cancer) and their sub-types. Thus, developing such deep learning method, which can successfully extract meaningful features from various breast cancer sub-types, is of current research interest. In this paper, we develop dual stage (unsupervised pre-training and supervised fine-tuning) neural network architecture termed AFExNet based on adversarial auto-encoder (AAE) to extract features from high dimensional genetic data. We evaluated the performance of our model through twelve different supervised classifiers to verify the usefulness of the new features using public RNA-Seq dataset of breast cancer. AFExNet provides consistent results in all performance metrics across twelve different classifiers which makes our model classifier independent. We also develop a method named 'TopGene' to find highly weighted genes from the latent space which could be useful for finding cancer bio-markers. Put together, AFExNet has great potential for biological data to accurately and effectively extract features. Our work is fully reproducible and source code can be downloaded from Github: https://github.com/NeuroSyd/breast-cancer-sub-types.
Asunto(s)
Neoplasias de la Mama , Algoritmos , Neoplasias de la Mama/diagnóstico , Neoplasias de la Mama/genética , Análisis por Conglomerados , Femenino , Humanos , Redes Neurales de la Computación , Programas InformáticosRESUMEN
Artificial intelligence (AI) and health sensory data-fusion hold the potential to automate many laborious and time-consuming processes in hospitals or ambulatory settings, e.g. home monitoring and telehealth. One such unmet challenge is rapid and accurate epileptic seizure annotation. An accurate and automatic approach can provide an alternative way to label seizures in epilepsy or deliver a substitute for inaccurate patient self-reports. Multimodal sensory fusion is believed to provide an avenue to improve the performance of AI systems in seizure identification. We propose a state-of-the-art performing AI system that combines electroencephalogram (EEG) and electrocardiogram (ECG) for seizure identification, tested on clinical data with early evidence demonstrating generalization across hospitals. The model was trained and validated on the publicly available Temple University Hospital (TUH) dataset. To evaluate performance in a clinical setting, we conducted non-patient-specific pseudo-prospective inference tests on three out-of-distribution datasets, including EPILEPSIAE (30 patients) and the Royal Prince Alfred Hospital (RPAH) in Sydney, Australia (31 neurologists-shortlisted patients and 30 randomly selected). Our multimodal approach improves the area under the receiver operating characteristic curve (AUC-ROC) by an average margin of 6.71% and 14.42% for deep learning techniques using EEG-only and ECG-only, respectively. Our model's state-of-the-art performance and robustness to out-of-distribution datasets show the accuracy and efficiency necessary to improve epilepsy diagnoses. To the best of our knowledge, this is the first pseudo-prospective study of an AI system combining EEG and ECG modalities for automatic seizure annotation achieved with fusion of two deep learning networks.
Asunto(s)
Inteligencia Artificial , Epilepsia , Electroencefalografía/métodos , Epilepsia/diagnóstico , Humanos , Estudios Prospectivos , Convulsiones/diagnósticoRESUMEN
We examined the white matter of patients with and without focal to bilateral tonic-clonic seizures (FBTCS), and control participants. A neural network based tract segmentation model (Tractseg) was used to isolate tract-specific, track-weighted tensor-based measurements from the tracts of interest. We compared the group differences in the track-weighted tensor-based measurements derived from whole and hemispheric tracts. We identified several regions that displayed significantly altered white matter in patients with focal epilepsy compared to controls. Furthermore, patients without FBTCS showed significantly increased white matter disruption in the inferior fronto-occipital fascicle and the striato-occipital tract. In contrast, the track-weighted tensor-based measurements from the FBTCS cohort exhibited a stronger resemblance to the healthy controls (compared to the non-FBTCS group). Our findings revealed marked alterations in a range of subcortical tracts considered critical in the genesis of seizures in focal epilepsy. Our novel application of tract-specific, track-weighted tensor-based measurements to a new clinical dataset aided the elucidation of specific tracts that may act as a predictive biomarker to distinguish patients likely to develop FBTCS.
RESUMEN
The majority of studies for automatic epileptic seizure (ictal) detection are based on electroencephalogram (EEG) data, but electrocardiogram (ECG) presents a simpler and more wearable alternative for long-term ambulatory monitoring. To assess the performance of EEG and ECG signals, AI systems offer a promising way forward for developing high performing models in securing both a reasonable sensitivity and specificity. There are crucial needs for these AI systems to be developed with more clinical relevance and inference generalization. In this work, we implement an ECG-specific convolutional neural network (CNN) model with residual layers and an EEG-specific convolutional long short-term memory (ConvLSTM) model. We trained, validated, and tested these models on a publicly accessible Temple University Hospital (TUH) dataset for reproducibility and performed a non-patient-specific inference-only test on patient EEG and ECG data of The Royal Prince Alfred Hospital (RPAH) in Sydney, Australia. We selected 31 adult patients to balance groups with the following seizure types: generalized, frontal, frontotemporal, temporal, parietal, and unspecific focal epilepsy. Our tests on both EEG and ECG of these patients achieve an AUC score of 0.75. Our results show ECG outperforms EEG with an average improvement of 0.21 and 0.11 AUC score in patients with frontal and parietal focal seizures, respectively.Clinical relevance-Prior research has demonstrated the value of using ECG for seizure documentation. It is believed that specific epileptic foci (seizure origin) may involve network inputs to the autonomic nervous system. Our result indicates that ECG could outperform EEG for individuals with specific seizure origin, particularly in the frontal and parietal lobes.