Paper deep dive
Light-Adapted Electroretinogram and Oscillatory Potentials (LEOPs) Dataset for Autism Spectrum Disorder and Typically Developing Individuals
Paul A. Constable, Dorothy A. Thompson, Irene O. Lee, Lynne Loh, Aleksei Zhdanov, Mikhail Kulyabin, Andreas Maier
Intelligence
Status: succeeded | Model: Gemma-4-26B-A4B | Prompt: intel-v1 | Confidence: 98%
Last extracted: 4/27/2026, 3:20:23 AM
Summary
The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides light-adapted electroretinogram (ERG) and Oscillatory Potentials (OPs) waveforms for three populations: typically developing Controls, Autism Spectrum Disorder (ASD), and ASD + ADHD. The dataset includes 5309 single flash ERG and 4434 OPs waveforms recorded using the RETeval device. It features three protocols (9-step, 2-step, and LA3 ISCEV standard) and provides detailed metadata including participant demographics, iris color, electrode position images, and clinical scores (FSIQ, ASD severity, CARS). The data is structured for machine learning applications, specifically for identifying biomarkers in neurodevelopmental disorders.
Entities (9)
Relation Signals (5)
LEOPs Dataset â containswaveformsfor â Autism Spectrum Disorder
confidence 100% · The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides... waveforms for... Autism Spectrum Disorder (ASD)
LEOPs Dataset â containswaveformsfor â Attention Deficit Hyperactivity Disorder
confidence 100% · The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides... waveforms for... ASD + Attention Deficit Hyperactivity Disorder (ADHD)
LEOPs Dataset â includesmetadata â Iris color
confidence 100% · The LEOPs dataset includes... alongside iris color, electrode position with image files
Flinders University â issiteof â LEOPs Dataset Collection
confidence 100% · Data were collected across two sites (Flinders University, Australia and University College London, UK).
RETeval â usedtorecord â Electroretinogram
confidence 100% · The ERGs were recorded... using the handheld RETeval device
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides light-adapted (LA) electroretinogram (ERG) and Oscillatory Potentials (OPs) waveforms for typically developing Control, Autism Spectrum Disorder (ASD) and ASD + Attention Deficit Hyperactivity Disorder (ADHD) childhood and adolescent populations. The ERGs were recorded in the Right And Left eyes with skin electrodes using the handheld RETeval device at two sites in Australia and the United Kingdom. The LEOPs dataset includes 5309 single flash ERG and 4434 OPs waveforms as well as images selected from each participant showing the position of the skin electrode. The LEOPs dataset is constructed from recordings using a 9 step randomized flash series from $-0.37$ to $1.20$~$Td.s$, a 2 step at 113 and 446 $Td.s$ flash strengths (2500 Control, 1730 ASD and 451 ASD + ADHD samples), as well as the $85$~$Td.s$ (Light Adapted 3 $cd.s.m^{-2}$ (LA3)) equivalent International Society of Clinical Electrophysiology of Vision (ISCEV) Standard flash with 435 Control, 176 ASD and 37 ASD + ADHD waveform samples. Code for the stimulus is provided along with participant demographics, date and time of testing, and where available diagnostic scores for the ASD and ASD + ADHD groups, alongside iris color, electrode position with image files and time domain values for the ERG and summed values for the OPs. The repository contains excel file, exported JSON files on the patient level that are more suitable for machine learning tasks, images of electrode position for each recording and the protocol files for use with the RETeval.
Tags
Links
- Source: https://arxiv.org/abs/2604.16981v1
- Canonical: https://arxiv.org/abs/2604.16981v1
Trouble viewing inline? Open PDF directly â
Full Text
45,884 characters extracted from source content.
Expand or collapse full text
Light-Adapted Electroretinogram and Oscillatory Potentials (LEOPs) Dataset for Autism Spectrum Disorder and Typically Developing Individuals Paul A. Constable 1 , Dorothy A. Thompson 2,3 , Irene O. Lee 4 , Lynne Loh 1 , Aleksei Zhdanov 5 , Mikhail Kulyabin 5 , and Andreas Maier 6 1 Flinders University, College of Nursing and Health Sciences, Caring Futures Institute, Adelaide, Australia 2 The Tony Kriss Visual Electrophysiology Unit, Clinical and Academic Department of Ophthalmology, Great Ormond Street Hospital for Children NHS Trust, London, United Kingdom 3 UCL Great Ormond Street Institute of Child Health, University College London, London, United Kingdom 4 Behavioural and Brain Sciences Unit, Population Policy and Practice Programme, UCL Great Ormond Street Institute of Child Health, University College London, London, United Kingdom 5 Visiomed.AI, Moscow, Russia 6 Pattern Recognition Lab, Friedrich-Alexander-Universit Ì at Erlangen-N Ì urnberg, 91058 Erlangen, Germany ABSTRACT The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides light-adapted (LA) electroretinogram (ERG) and Oscillatory Potentials (OPs) waveforms for typically developing Control, Autism Spectrum Disorder (ASD) and ASD + Attention Deficit Hyperactivity Disorder (ADHD) childhood and adolescent populations. The ERGs were recorded in the Right And Left eyes with skin electrodes using the handheld RETeval device at two sites in Australia and the United Kingdom. The LEOPs dataset includes 5309 single flash ERG and 4434 OPs waveforms as well as images selected from each participant showing the position of the skin electrode. The LEOPs dataset is constructed from recordings using a 9 step randomized flash series from â0.37to1.20 Td.s, a 2 step at 113 and 446Td.sflash strengths (2500 Control, 1730 ASD and 451 ASD + ADHD samples), as well as the85 Td.s(Light Adapted 3cd.s.m â2 (LA3)) equivalent International Society of Clinical Electrophysiology of Vision (ISCEV) Standard flash with 435 Control, 176 ASD and 37 ASD + ADHD waveform samples. Code for the stimulus is provided along with participant demographics, date and time of testing, and where available diagnostic scores for the ASD and ASD + ADHD groups, alongside iris color, electrode position with image files and time domain values for the ERG and summed values for the OPs. The repository contains excel file, exported JSON files on the patient level that are more suitable for machine learning tasks, images of electrode position for each recording and the protocol files for use with the RETeval. Background & Summary The LEOPs dataset 1 provides light adapted (LA) electroretinogram (ERG) and Oscillatory Potentials (OPs) averaged waveforms for the (right and left eyes with replicates) in children and adolescents diagnosed with autism spectrum disorder (ASD), ASD + Attention Deficit Hyperactivity Disorder (ADHD) and typically developing Control individuals. The full-field ERG is a biological signal recorded from the retina in response to a brief flash of light. The resulting waveform is generated from pho- toreceptors, bipolar, glial, horizontal, amacrine and ganglion cells 2â5 . The International Society for Clinical Electrophysiology of Vision (ISCEV) recommends a series of standard ERGs 6 and supports the availability of normative databases to be publicly available to integrate into a harmonized reference dataset library 7 . The LEOPs dataset provides ISCEV standard light-adapted 3cd.s.m â2 (LA3) ERGs for ASD and Control groups as well as a nine-flash strength sequence that describes the light-adapted stimulus response function known as the photopic hill 8 . In addition, where available the OPs waveforms are provided to explore the high frequency components of the ERG waveform 9 . The LEOPs dataset provides potential to use the Control waveforms as part of a normative database or reference comparison group. The ASD group dataset could be used in conjunction with other biological or psychophysical biomarkers in studies exploring the classification of ASD from other groups using AI 10 . The OPs waveforms provide an additional resource with which to explore the changes in the OPs over a wide 9 step flash series. In addition, the LEOPs dataset includes the standard LA3 response that is part of the ISCEV recommended flash series 2 . Importantly, images of the skin electrode position are provided that are linked to each subject and recording so that future analyses could be made to ascertain the impact of skin electrode position of the recorded signal 11 . Iris color data is also included in the LEOPs dataset to compensate for the impact of iris arXiv:2604.16981v1 [physics.med-ph] 18 Apr 2026 color/ pigment on the ERG amplitude 12 . Time of day and date for each recording is also included to potentially study circadian variations in the ERG waveform 13, 14 . The LEOPs dataset as recorded using the RETeval (LKC Technologies, Germantown, MD, USA) with skin electrodes. Therefore, the LEOPs dataset would provide a comparative dataset for ERGs recorded with other electrode types such as thread, gold foil or contact lens 15 . Prior publications based on these data included attempts to classify ASD from Control and ADHD groups using machine learning using spectral domain analysis including Discrete Wavelet Transform, Variable Frequency Complex Demodulation and Continuous Wavelet Transform that provide moderate to high classification for neurodevelopmental groups with > 80% accuracy 16â19 . Time series-based analysis to identify key parts of the ERG waveform to inform classification models 20 . The application of a gated multi-layer perceptron analysis 21 yielded a classification accuracy of 89.7%. Finally short-time Fourier transforms have been used to generate features for deep learning to classify these groups achieving an AUC of 0.89 22 . The averaged waveforms have been used to generate synthetic ERG waveforms to bolster the training and testing datasets for ML models as well as provide an increase in sample size to balance unbalanced datasets 23, 24 . Further analyses have incorporated aspects of Functional Data Analysis and Bayesian statistics to model the rising portion of the b-wave to discriminate changes in the dynamic profile of the ERG waveform between the ASD and Control groups 25â27 . The waveforms have also been used as an additional marker for psychophysical markers in ASD and typically developing children 28 . Whilst the LEOPs dataset comprises ASD, ASD + ADHD, and Control groups further applications could be used to differentiate between the wider neurodevelopmental spectrum including ADHD 29â32 and drug interactions with the retina 33â35 . Whilst the main features of the LEOPs dataset demonstrate group differences have been observed in the b-wave of the ERG, studies focusing on the later Photopic Negative Response (PhNR) that is driven by ganglion cells could also utilize these data for retinal and neurodevelopmental disorders 4, 15, 36 . The available protocol files have also been used in one prior study investigating the ERG in adults with ASD 37 and could be used for related neurological disorders in which the LA-ERG that form the photopic hill are atypical such as Parkinsonâs Disease, schizophrenia and Bipolar disorder 35, 38â40 . The LEOPs dataset also includes the OPs waveforms that are the focus of novel analytical methods including the Hilbert Transform 41 and will provide data with which to explore ways of modeling these waveforms. The photopic hill data series may also be beneficial to expanding the mathematical modeling of this stimulus-response function 42, 43 . The 9 step series provides a resource with which to model the a-wave kinetics across an extended flash series 44 and the PhNR that also has different modalities in which to assess 45, 46 . The LEOPs dataset will provide additional resources for the classification of neurodevelopmental disorders that use functional and/or structural biomarkers 38, 47â50 that span Visual Transformers 51 , AI modeling of pediatric ERGs 52 to support the future development of AI driven tools for mental health 53 . Related datasets include in time series values for ASD and Control groups and ISCEV standard ERGs recorded with the RETeval but no raw waveforms 54 . Pattern ERG waveforms have been made publicly available for assessment of macular function 55 . Normative ranges but no waveforms for skin electrodes in a pediatric population are also published 56 . Reference ranges in a Chinese childhood population for the ISCEV standard series using the RETeval is also available but without waveform, data 57 . Reference limits for the full field dark and light adapted ERGs in 407 participants with three electrode types across two sites is available 58 . Synthetically generated ERG waveforms from four flash strengths of ASD and Control groups is available at IEEE DataPort 59 . In contrast to current publicly available datasets the LEOPs dataset provides averaged waveforms in JSON and excel files for both ERG and OPs, collected at two sites with detailed demographic data for Control, ASD and ASD + ADHD populations. Psychological test data for the ASD and and ASD + ADHD groups is also reported where available. The LEOPs dataset has provided data in prior publications investigating the ERG as a marker for ASD 16, 18â27, 30, 31, 36, 51, 59, 60 . Methods Detailed descriptions of data collection methods and group characteristics have been previously reported 19, 29, 60 . All stimulus frequencies were at1.96294 Hzand the date and time of day for each recording is provided along with site, sex, age, iris color as reported by the RETeval based on grey scale ratio of the pupil to a3 mmstrip on the iris at the 3 and 9 oâclock position from the pupil margin. The OPs waveforms are provided for most waveforms and were extracted post-hoc from the averaged ERG waveform through band-pass digital filtering between85â195 Hz. (1) The 9-step which consists of 9 randomized flash strengths recorded on a white40 cd.m â2 background with 60 averages to generate the reported averaged ERG waveform. Flash strengths included12,21,35,70,113,178,251,356and446 Td.son a 1130Td.s(40 cd.m â2 ) white background with 60 averages. The flash strengths have an equivalentlog cd.m â2 values ofâ0.37,â0.12,0.11,0.40,0.60,0.80,0.95,1.11and 1.20based on a6 mmpupil diameter. Note waveforms have a baseline of20 msand recording interval0â100 ms. (2) The 2-step protocol used only two flash strengths (113 Td.sand446 Td.s) on a40 cd.m â2 white background with 30 averages were employed to generate the reported average ERG waveform. Note the pre-stimulus baseline for these recordings is extended to 50 msand the recording interval is longer and from0to170 ms. (3) The LA3 (85 Td.s) protocol was run as part of the 9-step protocol randomized series with ERG waveforms recorded at the end of the series. For the LA3 recordings were made on a on a 2/12 30 cd.m â2 (848Td.s) white background with 30 averages. Additional Control waveforms have been added from the reference dataset at site 1. Clinical psychological test data for the ASD and ASD + ADHD groups are provided, including the Full scale IQ, Autism severity score based on the methods of Gotham et al., (2009) 61 and Childhood Autism Rating Scale (CARS) 62 . All ASD or ASD + ADHD participants met diagnostic criteria according to DSM-IV-TR 63 or DSM-V 64 which was supported either by clinical report including Autism Diagnostic Observational Schedule-2 (ADOS) 65 and classed as high diagnostic confidence provided by hospital based assessment; or by community based assessment supported by CARS 62 â classed as medium confidence or without CARS - classed as low confidence. Participants with co-occurring ASD and ADHD were included in the dataset for completeness. Ethics The studies were conducted in accordance with the Declaration of Helsinki at sites based in Australia (Flinders University) and the United Kingdom (University College London). Ethics approvals were provided for the studies from the Southern Adelaide Clinical Human Research Ethics Committee (Approval Code: 318.16) and the Flinders University Human Research and Ethics Committee (Approval Codes: 4606 and 7180) and the Southeast Scotland Research Ethics Committee in the UK (Approval Code:18/S/0008). Data records The LEOPs dataset 1 comprises 5309 ERG averaged waveforms from 253 participants (157 TD, 75 ASD, 21 ASD + ADHD), provided in three complementary formats: an Excel spreadsheet, structured JSON files, and eye image files. Data were collected across two sites (Flinders University, Australia and University College London, UK). Excel spreadsheet The fileLEOPs_dataset.xlsxcontains five worksheets. Three waveform sheets (9_step,2_step, andLA3) store, for each waveform, the per-recording metadata and the reported averaged ERG and, where available, OPs waveform timeâ amplitude signals in paired columns (ms,ÎŒV). The9_stepsheet contains 4,246 waveforms from 173 participants with 235 data points per waveform (20 msbaseline,0â100 msrecording interval). The2_stepsheet contains 415 waveforms from 61 participants with 430 data points per waveform (50 msbaseline,0â170 msrecording interval). TheLA3sheet contains 648 waveforms from 217 participants with 235 data points per waveform. Two participant sheets (Participants 9 and 2andparticipants LA3) provide a tabular view of the per-waveform metadata including demographics, stimulus parameters, and time-domain features. Each waveform is identified by a unique waveform identifier (e.g.a100-1) comprising the participant identifier and a sequence number. Per-waveform metadata includes: participant group (Control= 0, ASD or ASD + ADHD = 1); diagnosis confidence (1 = high, 2 = medium, 3 = low); recording site; age at test (years); sex at birth (male = 0, female = 1); electrode vertical position (+1, 0,â1,â2 relative to the manufacturerâs recommended height of 2 m below the lower lid); medication status; Full Scale IQ; ASD severity score 61 ; CARS score 62 ; iris color ratio; test date and time; tested eye; and flash strength in both Td.sandlog cd.s.m â2 . Time-domain ERG features provided are the a-wave time to peak and amplitude, the b-wave time to peak and amplitude (measured from the a-wave minimum), and the summed OPs amplitude and time where available. Of the 5309 waveforms, 4,434 (83.5%) include OPs waveforms extracted through band-pass digital filtering (85â195 Hz). JSON files To facilitate machine learning and deep learning workflows, the LEOPs dataset includes structured JSON files in the database/jsons/directory, with one file per participant (253 files total). Each JSON file aggregates all recordings for a participant across all protocols. The structure of each JSON file is as follows: "participant_id": string, "demographics": "category": int, "group": string, "diagnosis_strength": int|null, "site": int, "sex": int, "medication": int, "fsiq": int|null, "asd_severity": int|null, "cars": float|null , "eye_images": "right": string, "left": string |null, "recordings": [ "wave_id": string, "protocol": string, "age": float, "notes": string|null, "test_date": string, "test_time": string, "test_eye": string, "electrode_position": int, "iris": float, "stimulus": "flash_tds": float, "flash_cd": float, "frequency_hz": float , "features": "a_time_ms": float, "a_amp_uv": float, "b_time_ms": float, "b_amp_uv": float, "op_sum_amp_uv": float|null, "op_sum_time_ms": float|null , "erg_waveform": "time_ms": array[float], "amplitude_uv": array[float] , "op_waveform": "time_ms": array[float], "amplitude_uv": array[float] |null ] 3/12 Position: +1 (a) Position: 0 Position: 1Position: 2 (b) Figure 1. (a) Eye image files from the RETeval are provided as part of the dataset. Electrode position is a key factor in determining the amplitude of the recorded signal. The manufacturer recommends to place the electrode2 mmbelow the lower lid. Images have been subjectively graded as +1 =1 mmabove the recommended height; 0 = at the recommended height;â1 or â2⌠1or2 mmbelow the recommended height respectively. (b) Iris color is calculated by the grey scale ratio of a3 mmstrip besides the pupil indicated by the white lines to the black pupil. The demographics object contains participant-level fields that are constant across all recordings: numeric category (Control= 0, ASD or ASD + ADHD = 1), diagnostic group (Control, ASD, or ASD + ADHD), diagnosis confidence (1 = high, 2 = medium, 3 = low, or null if unavailable), recording site, sex (0 = male, 1 = female), medication status, and where available clinical scores (FSIQ, ASD severity, CARS). Theeye_imagesobject links to the right and left eye image filenames, or isnullif unavailable. Therecordingsarray contains one entry per waveform, each with recording-level metadata that may vary across sessions: age (years), notes, test date and time, tested eye, electrode position (+1, 0,â1,â2 relative to the manufacturerâs recommended height, graded per eye from the corresponding eye image), iris color, stimulus parameters (flash strength inTd.s andlog cd.s.m â2 , frequency inHz), time-domain features (a-wave and b-wave timing and amplitude, summed OPs), and the raw ERG and OPs waveforms as paired timeâamplitude arrays. Theop_waveformfield isnullfor recordings without oscillatory potential data. Conversion from Td.s to cd.s.m â2 was performed assuming a 6 m pupil diameter. Eye images A total of 558 PNG eye images (278 right eye, 279 left eye) are provided in theImagesERG/directory, covering 244 of 253 participants. Each image shows the position of the RETeval skin electrode relative to the lower eyelid and has been subjectively graded for electrode position as shown in Fig. 1. Image filenames follow the conventionparticipant_id_ RightEye|LeftEye.png and are cross-referenced in both the Excel and JSON files. Table 1. Summary of dataset parameters as mean± SEM for each group and data set series. OPs = Oscillatory Potentials; TD = Typically Developing. Nine and Two Step protocols NAgeSexIrisa-timea-ampb-timeb-ampÎŁOPs-tÎŁOPs-a (yr)M : F %(ms)(ÎŒV )(ms)(ÎŒV )(ms)(ÎŒV ) ASD173013.5± 0.180 : 201.235± 0.00312.11± 0.04â6.37± 0.0827.23± 0.0824.32± 0.26132.05± 0.6422.80± 0.29 ASD+ADHD43114.5± 0.268 : 321.182± 0.00411.96± 0.09â7.37± 0.1527.27± 0.1528.20± 0.55132.07± 1.1427.71± 0.57 TD250013.5± 0.148 : 521.258± 0.00312.07± 0.03â7.33± 0.0727.00± 0.0627.86± 0.22132.84± 0.4826.29± 0.22 LA3 ISCEV Standard NAgeSexIrisa-timea-ampb-timeb-ampÎŁOPs-tÎŁOPs-a (yr)M : F %(ms)(ÎŒV )(ms)(ÎŒV )(ms)(ÎŒV ) ASD17613.7± 0.380 : 201.241± 0.00811.48± 0.08â6.80± 0.2228.44± 0.0928.93± 0.77131.07± 2.2227.13± 0.83 ASD+ADHD3715.2± 0.762 : 381.196± 0.01612.07± 0.35â7.77± 0.4027.94± 0.1734.69± 2.05125.07± 3.5431.91± 2.14 TD43513.5± 0.337 : 631.232± 0.00611.58± 0.05â7.17± 0.1127.91± 0.0432.30± 0.48131.44± 1.6927.37± 0.60 4/12 Data overview Table 1 summarizes the LEOPs dataset parameters for each group, presented as two sub-tables: one for the combined 9-step and 2-step protocols and one for the LA3 ISCEV standard protocol, which was collected separately and includes additional Control participants from the Flinders University reference dataset. The total number of waveforms (N) includes replicates within eyes across all flash strengths and excludes OPs waveforms. Mean±SEM values are provided for age, iris color ratio, a-wave and b-wave time-domain parameters, and summed OPs amplitudes and implicit times. The sex ratio differs markedly between groups: the ASD group shows a strong male predominance (80 : 20 M : F), consistent with the higher prevalence of ASD diagnoses in males 63 , whereas the Control group has a more balanced distribution (49 : 51 in the 9/2-step data). Iris color is reported as the grey-scale ratio of iris to pupil reflectance measured by the RETeval; values 1.1 indicate lighter irides and higher values 1.5 indicate darker irides (Fig. 1). The ASD group shows a reduced mean b-wave amplitude (24.32± 0.26ÎŒV) compared to TD (27.86± 0.22ÎŒV) across the 9-step and 2-step protocols, consistent with previous reports of attenuated retinal responses in ASD 60 . The ASD + ADHD group shows intermediate or comparable values to the Control group, though with wider SEM reflecting the smaller sample size (n=431 ASD + ADHD vs 2500 for Control). 20020406080100 10 0 10 20 30 Amplitude (ÎŒV) (a) 12 Td·s (-0.37 log cd·s·m 2 ) Control (n=263) ASD+ADHD (n=43) ASD (n=189) 20020406080100 10 0 10 20 30 (b) 21 Td·s (-0.12 log cd·s·m 2 ) Control (n=237) ASD+ADHD (n=39) ASD (n=191) 20020406080100 10 0 10 20 30 (c) 35 Td·s (+0.11 log cd·s·m 2 ) Control (n=236) ASD+ADHD (n=39) ASD (n=184) 20020406080100 10 0 10 20 30 Amplitude (ÎŒV) (d) 70 Td·s (+0.40 log cd·s·m 2 ) Control (n=238) ASD+ADHD (n=42) ASD (n=189) 20020406080100 10 0 10 20 30 (e) 113 Td·s (+0.60 log cd·s·m 2 ) Control (n=238) ASD+ADHD (n=43) ASD (n=185) 20020406080100 10 0 10 20 30 (f) 178 Td·s (+0.80 log cd·s·m 2 ) Control (n=233) ASD+ADHD (n=41) ASD (n=198) 20020406080100 Time (ms) 10 0 10 20 30 Amplitude (ÎŒV) (g) 251 Td·s (+0.95 log cd·s·m 2 ) Control (n=235) ASD+ADHD (n=42) ASD (n=187) 20020406080100 Time (ms) 10 0 10 20 30 (h) 356 Td·s (+1.11 log cd·s·m 2 ) Control (n=240) ASD+ADHD (n=39) ASD (n=194) 20020406080100 Time (ms) 10 0 10 20 30 (i) 446 Td·s (+1.20 log cd·s·m 2 ) Control (n=245) ASD+ADHD (n=43) ASD (n=193) Figure 2. Mean±1 SD light-adapted ERG waveforms for the 9-step protocol across three groups: Control (blue), ASD (red), and ASD + ADHD (orange). Each panel shows the grand-average waveform (solid line) with shaded±1 SD region for a single flash strength. (a) 12 Td.s. (b) 21 Td.s. (c) 35 Td.s. (d) 70 Td.s. (e) 113 Td.s. (f) 178 Td.s. (g) 251 Td.s. (h) 356 Td.s. (i) 446 Td.s. The dashed vertical line indicates flash onset (0 ms); the 20 ms pre-stimulus baseline precedes it. The n values denote the number of individual waveforms (including both eyes where available) contributing to each average. 5/12 Fig. 2 presents the mean ERG waveforms with±1 SD shading for the 9-step protocol across all nine flash strengths, grouped by category (Control, ASD, ASD + ADHD). The characteristic photopic hill is evident as b-wave amplitude increases from12to70 Td.s, plateaus around113â178 Td.s, and then decreases at251â446 Td.s. The ASD group shows a reduced b-wave amplitude relative to the Control group at most flash strengths, consistent with prior findings 60 . The ASD + ADHD group follows a similar pattern with greater variability reflecting the smaller sample size. 20020406080100 10 5 0 5 10 Amplitude (ÎŒV) (a) 12 Td·s (-0.37 log cd·s·m 2 ) Control (n=246) ASD+ADHD (n=43) ASD (n=142) 20020406080100 10 5 0 5 10 (b) 21 Td·s (-0.12 log cd·s·m 2 ) Control (n=221) ASD+ADHD (n=39) ASD (n=145) 20020406080100 10 5 0 5 10 (c) 35 Td·s (+0.11 log cd·s·m 2 ) Control (n=219) ASD+ADHD (n=39) ASD (n=138) 20020406080100 10 5 0 5 10 Amplitude (ÎŒV) (d) 70 Td·s (+0.40 log cd·s·m 2 ) Control (n=220) ASD+ADHD (n=42) ASD (n=144) 20020406080100 10 5 0 5 10 (e) 113 Td·s (+0.60 log cd·s·m 2 ) Control (n=219) ASD+ADHD (n=42) ASD (n=139) 20020406080100 10 5 0 5 10 (f) 178 Td·s (+0.80 log cd·s·m 2 ) Control (n=216) ASD+ADHD (n=41) ASD (n=147) 20020406080100 Time (ms) 10 5 0 5 10 Amplitude (ÎŒV) (g) 251 Td·s (+0.95 log cd·s·m 2 ) Control (n=217) ASD+ADHD (n=42) ASD (n=140) 20020406080100 Time (ms) 10 5 0 5 10 (h) 356 Td·s (+1.11 log cd·s·m 2 ) Control (n=222) ASD+ADHD (n=39) ASD (n=143) 20020406080100 Time (ms) 10 5 0 5 10 (i) 446 Td·s (+1.20 log cd·s·m 2 ) Control (n=225) ASD+ADHD (n=43) ASD (n=143) Figure 3. Mean± 1 SD Oscillatory Potentials (OPs) waveforms for the 9-step protocol across three groups: Control (blue), ASD (red), and ASD + ADHD (orange). OPs were extracted from the averaged ERG waveform by band-pass digital filtering (85â195 Hz). Each panel shows the grand-average OPs waveform (solid line) with shaded±1 SD region for a single flash strength. (a) 12 Td.s. (b) 21 Td.s. (c) 35 Td.s. (d) 70 Td.s. (e) 113 Td.s. (f) 178 Td.s. (g) 251 Td.s. (h) 356 Td.s. (i) 446 Td.s. The n values denote the number of individual OPs waveforms contributing to each average. Fig. 3 presents the mean OPs waveforms for the 9-step protocol. The OPs represent high-frequency components of the ERG waveform generated primarily by amacrine cell interactions in the inner retina. The OPs amplitude increases with flash strength across the photopic hill, with the most prominent oscillations visible at the higher flash strengths (113â446 Td.s). Fig. 4 shows the mean ERG waveforms for the 2-step protocol, which employed a longer recording window (50 msbaseline, 0â170 mspost-stimulus) with two flash strengths and 30 averages. The ASD group has a notably small sample size (n=10 waveforms) in this protocol subset, resulting in wider SD bands. OPs are also illustrated for this dataset, with OPs also available for the 9-step protocol. 6/12 50050100150 Time (ms) 30 20 10 0 10 20 30 Amplitude (ÎŒV) (a) 113 Td·s (+0.60 log cd·s·m 2 ) Control (n=163) ASD+ADHD (n=28) ASD (n=10) 50050100150 Time (ms) 30 20 10 0 10 20 30 (b) 446 Td·s (+1.20 log cd·s·m 2 ) Control (n=172) ASD+ADHD (n=32) ASD (n=10) Figure 4. Mean±1 SD light-adapted ERG waveforms for the 2-step protocol across three groups: Control (blue), ASD (red), and ASD + ADHD (orange). (a) 113 Td.s. (b) 446 Td.s. The recording window extends fromâ50 to 170 ms, capturing later waveform morphology beyond the b-wave. Oscillatory Potentials waveforms are also shown for the three groups. Thenvalues denote the number of individual waveforms contributing to each average. 50050100150 Time (ms) 7.5 5.0 2.5 0.0 2.5 5.0 7.5 Amplitude (ÎŒV) (a) 113 Td·s (+0.60 log cd·s·m 2 ) Control (n=163) ASD+ADHD (n=28) ASD (n=10) 50050100150 Time (ms) 7.5 5.0 2.5 0.0 2.5 5.0 7.5 (b) 446 Td·s (+1.20 log cd·s·m 2 ) Control (n=172) ASD+ADHD (n=32) ASD (n=10) Figure 5. Mean± 1 SD Oscillatory Potentials (OPs) waveforms for the 2-step protocol across three groups: Control (blue), ASD (red), and ASD + ADHD (orange). (a) 113 Td.s. (b) 446 Td.s. The extended recording window (50 ms baseline, 0â170 ms). The n values denote the number of individual OPs waveforms contributing to each average. 20020406080100 Time (ms) 10 0 10 20 30 Amplitude (ÎŒV) 85 Td·s (+0.48 log cd·s·m 2 ) Control (n=435) ASD+ADHD (n=37) ASD (n=176) Figure 6. Mean± 1 SD light-adapted ERG waveform for the ISCEV standard LA3 protocol (85 Td.s) across three groups: Control (blue, n=435), ASD (red, n=176), and ASD + ADHD (orange, n=37). The recording parameters include a (20 ms baseline, 0â100 ms recording interval, with 30 averages). 7/12 Fig. 5 shows the corresponding OPs waveforms for the 2-step protocol. The Control group (n=335) shows the clearest OPs morphology, while the ASD group (n=20) has wider variability due to the smaller sample size. Fig. 6 presents the ISCEV standard LA3 (85 Td.s) ERG waveform. This protocol subset includes additional Control waveforms in young adults from the Flinders University reference dataset, yielding the largest sample size in the dataset. Technical Validation Signal quality was assured at two levels. First, during acquisition the RETeval device applied automatic artifact rejection: individual waveform traces containing blinks or other artifacts that fell within the upper or lower quartile of the running average were excluded before computing the reported averaged ERG waveform. Each averaged waveform therefore represents 30â60 artifact-free traces per eye. Second, a post-hoc visual inspection was performed to remove any remaining waveforms that exhibited excessive noise or large baseline offsets. Data completeness. All 5309 ERG waveforms in the dataset contain complete timeâamplitude arrays and time-domain features (a-wave and b-wave amplitude and implicit time). OPs waveforms are available for 4,434 (83.5%) of recordings, with the remaining 875 waveforms lacking OPs data due to recording protocol differences. Eye images are provided for 244 of 253 participants (96.4%). Eye images were excluded if the captured image occurred during a blink or eccentric fixation. Demographic consistency. Participant-level demographic fields (diagnostic group, sex, recording site, medication status, and clinical scores) were cross- validated across all waveform sheets to verify internal consistency. Automated checks confirmed that, for each participant, these fields are identical across all recordings in the 9-step, 2-step, and LA3 protocols. StimulusâResponse function. The mean b-wave amplitude for the Control group across the 9-step protocol follows the expected photopic hill profile 8 : amplitude increases from12.2± 4.4ÎŒVat12 Td.sto a plateau of33.5± 9.6ÎŒVat70â113 Td.s, before declining to 28.6± 8.6ÎŒVat446 Td.s(Fig. 7a). This physiologically expected non-monotonic stimulusâresponse relationship confirms that the stimulus calibration and recording parameters are consistent with established norms. The 2-step protocol (Fig. 7b) and LA3 protocol (Fig. 7c) show consistent b-wave amplitudes at their respective flash strengths, with group differences comparable to those observed in the 9-step data. 0.40.20.00.20.40.60.81.01.2 Flash strength (log cd·s·m 2 ) 10 15 20 25 30 35 b-wave amplitude (ÎŒV) (a) 9-step protocol Control (n240) ASD+ADHD (n41) ASD (n190) 0.60.70.80.91.01.11.2 Flash strength (log cd·s·m 2 ) 10 15 20 25 30 35 (b) 2-step protocol Control (n167) ASD+ADHD (n30) ASD (n10) 0.460.470.480.490.50 Flash strength (log cd·s·m 2 ) 10 15 20 25 30 35 (c) LA3 protocol Control (n435) ASD+ADHD (n37) ASD (n176) 12213570113178251356446113446 Flash strength (Td·s) 85 Figure 7. Stimulusâresponse function across all three protocols. Mean b-wave amplitude (± SEM) is plotted against flash strength inlog cd.s.m â2 (bottom axis) andTd.s(top axis) for each group: Control (blue circles), ASD (red squares), and ASD + ADHD (orange diamonds). (a) 9-step protocol: the non-monotonic photopic hill profile, with amplitude peaking near 70â113 Td.s before declining at higher flash strengths. (b) 2-step protocol (113 and 446 Td.s). (c) LA3 protocol (85 Td.s). Approximate sample sizes per group are indicated in the legend. Inter-ocular consistency. For participants with bilateral recordings, the b-wave amplitude showed strong inter-ocular correlation across all protocols and flash strengths (Fig. 8). In the 9-step protocol (Fig. 8a), Pearsonrranged from0.67to0.85(allp< 10 â20 ). Paired t-tests revealed no significant difference between right and left eyes at any flash strength (allp> 0.05), with mean inter-ocular 8/12 differences of less than1ÎŒV. Similar inter-ocular agreement was observed for the 2-step (Fig. 8b) and LA3 (Fig. 8c) protocols. This demonstrates good within-subject testâretest reliability of the recording setup across all stimulus conditions. Cross-site comparison. Data were collected at two sites: Flinders University, Adelaide, Australia (Site 1;n= 170 participants) and University College London, United Kingdom (Site 2;n= 82 participants). Within the Control group, b-wave amplitudes were significantly higher at Site 2 than Site 1 across all flash strengths (p< 0.005). For example, at113 Td.sthe mean b-wave amplitude was 29.9± 8.4ÎŒVat Site 1 versus35.5± 9.7ÎŒVat Site 2. These inter-site amplitude differences are consistent with known sources of variability in skin electrode ERG recordings, with the RETeval including differences in electrode position 11 , pigmentation 12 , fixation and waveform variability 66, 67 and operator variability 68 . Recording site is provided as a covariate in the dataset to enable users to account for this variability in downstream analyses. 10203040506070 Right eye b-wave (ÎŒV) 10 20 30 40 50 60 70 Left eye b-wave (ÎŒV) n = 1435 r = 0.871 = 0.32 ± 5.42 ÎŒV (a) 9-step protocol 2030405060 Right eye b-wave (ÎŒV) 20 30 40 50 60 n = 107 r = 0.738 = 0.65 ± 6.54 ÎŒV (b) 2-step protocol 102030405060 Right eye b-wave (ÎŒV) 10 20 30 40 50 60 n = 200 r = 0.808 = -0.65 ± 6.26 ÎŒV (c) LA3 protocol Figure 8. Inter-ocular consistency of b-wave amplitude across all three protocols. Each point represents the mean b-wave amplitude for a single participantâflash-strength combination, plotted as right eye versus left eye. The dashed grey line indicates perfect agreement (identity line) and the red line shows the linear regression fit. Pearson correlation (r), number of paired observations (n), and mean rightâleft difference (â) are annotated. (a) 9-step protocol. (b) 2-step protocol. (c) LA3 protocol. Data availability The LEOPs dataset available at https://data.mendeley.com/datasets/w3yx7hdds7. Code availability The code for loading, visualizing, and validating the LEOPs dataset and used to generate the results in this paper is available at https://github.com/MikhailKulyabin/LEOPs. References 1. Constable, P. et al. Light-adapted electroretinogram and oscillatory potentials (leops) dataset for autism spectrum disorder and typically developing individuals, 10.17632/w3yx7hdds7.1, (2026). [Data set]. 2. Robson, J. G., Saszik, S. M., Ahmed, J. & Frishman, L. J. Rod and cone contributions to the a-wave of the electroretinogram of the macaque. The J. physiology 547, 509â530, (2003). 3.Wachtmeister, L. Oscillatory potentials in the retina: what do they reveal. Prog. retinal eye research 17, 485â521, (1998). 4.Viswanathan, S., Frishman, L. J., Robson, J. G. & Walters, J. W. The photopic negative response of the flash electroretino- gram in primary open angle glaucoma. Investig. ophthalmology & visual science 42, 514â522, (2001). 5.Friedburg, C., Allen, C., Mason, P. & Lamb, T. D. Contribution of cone photoreceptors and post-receptoral mechanisms to the human photopic electroretinogram. The J. physiology 556, 819â834, (2004). 6.Robson, A. G. et al. Iscev standard for full-field clinical electroretinography (2022 update). Documenta Ophthalmol. 144, 165â177, (2022). 9/12 7.Hamilton, R. Clinical electrophysiology of visionâcommentary on current status and future prospects. Eye 35, 2341â2343, (2021). 8.Wali, N. & Leguire, L. E. The photopic hill: a new phenomenon of the light adapted electroretinogram. Documenta Ophthalmol. 80, 335â342, (1992). 9. Gauvin, M. et al. Assessing the contribution of the oscillatory potentials to the genesis of the photopic erg with the discrete wavelet transform. BioMed research international 2016, 2790194, (2016). 10.Patel, R., Jerskey, B. A., Shannon, J., Soares, N. & Fogler, J. M. Ai-enabled technologies and biomarker analysis for the early identification of autism and related neurodevelopmental disorders. Children 12, 1670, (2025). 11.Hobby, A. E. et al. Effect of varying skin surface electrode position on electroretinogram responses recorded using a handheld stimulating and recording system. Documenta Ophthalmol. 137, 79â86, (2018). 12. Al Abdlseaed, A., McTaggart, Y., Ramage, T., Hamilton, R. & McCulloch, D. L. Light-and dark-adapted electroretinograms (ergs) and ocular pigmentation: comparison of brown-and blue-eyed cohorts. Documenta ophthalmologica 121, 135â146, (2010). 13. Nozaki, S., Wakakura, M. & Ishikawa, S. Circadian rhythm of human electroretinogram. Jpn. journal ophthalmology 27, 346â352, (1983). 14. Hankins, M., Jones, R. & Ruddock, K. Diurnal variation in the b-wave implicit time of the human electroretinogram. Vis. neuroscience 15, 55â67, (1998). 15.Dubois, M.-A. et al. Validating skin electrodes: paving the way for non-invasive erg use in psychiatry. Prog. Neuro- Psychopharmacology Biol. Psychiatry 137, 111305, (2025). 16.Manjur, S. M. et al. Detecting autism spectrum disorder and attention deficit hyperactivity disorder using multimodal time-frequency analysis with machine learning using the electroretinogram from two flash strengths. J. Autism Dev. Disord. 55, 1365â1378, (2025). 17.Manjur, S. M. et al. Detecting autism spectrum disorder using spectral analysis of electroretinogram and machine learning: Preliminary results. In 2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 3435â3438 (IEEE, 2022). 18.Posada-Quintero, H. F. et al. Autism spectrum disorder detection using variable frequency complex demodulation of the electroretinogram. Res. Autism Spectr. Disord. 109, 102258, (2023). 19.Constable, P. A. et al. Spectral analysis of light-adapted electroretinograms in neurodevelopmental disorders: Classification with machine learning. Bioengineering 12, 15, (2024). 20. Chistiakov, S. et al. Time series classification of autism spectrum disorder using the light-adapted electroretinogram. Bioengineering 12, 951, (2025). 21. Kulyabin, M. et al. Attention to the electroretinogram: Gated multilayer perceptron for asd classification. IEEE access 12, 52352â52362, (2024). 22.Albasu, F. et al. Electroretinogram analysis using a short-time fourier transform and machine learning techniques. Bioengineering 11, 866, (2024). 23.Kulyabin, M. et al. Synthetic electroretinogram signal generation using a conditional generative adversarial network. Documenta Ophthalmol. 151, 161â177, (2025). 24. Kulyabin, M. et al. Generating synthetic light-adapted electroretinogram waveforms using artificial intelligence to improve classification of retinal conditions in under-represented populations. J. Ophthalmol. 2024, 1990419, (2024). 25.Brabec, M. et al. contour plot visualization of the light adapted electroretinogram using a generalized additive model. Documenta Ophthalmol. 1â11, (2026). 26.Brabec, M., Constable, P. A., Thompson, D. A. & Marmolejo-Ramos, F. Group comparisons of the individual electroretino- gram time trajectories for the ascending limb of the b-wave using a raw and registered time series. BMC Res. Notes 16, 238, (2023). 27.Brabec, M. et al. Remodeling the light-adapted electroretinogram using a bayesian statistical approach. BMC Res. Notes 18, 33, (2025). 28.Lee, I. O. et al. Global motion coherent deficits in individuals with autism spectrum disorder and their family members are associated with retinal function. Sci. Reports 15, 28249, (2025). 10/12 29.Lee, I. O. et al. The electroretinogram b-wave amplitude: a differential physiological measure for attention deficit hyperactivity disorder and autism spectrum disorder. J. neurodevelopmental disorders 14, 30, (2022). 30.Constable, P. A. et al. Discrete wavelet transform analysis of the electroretinogram in autism spectrum disorder and attention deficit hyperactivity disorder. Front. Neurosci. 16, 890461, (2022). 31.Manjur, S. M. et al. Spectral analysis of electroretinography to differentiate autism spectrum disorder and attention deficit hyperactivity disorder. In 2023 IEEE EMBS International Conference on Biomedical and Health Informatics (BHI), 1â4 (IEEE, 2023). 32.Dubois, M.-A. et al. Evaluation of electroretinography (erg) parameters as a biomarker for adhd. Prog. Neuro- Psychopharmacology Biol. Psychiatry 127, 110807, (2023). 33. Huang, Q. et al. Retinal gabaergic alterations in adults with autism spectrum disorder. J. Neurosci. 44, (2024). 34.Constable, P. A., Skuse, D. H., Thompson, D. A. & Lee, I. O. Brief report: effects of methylphenidate on the light adapted electroretinogram. Documenta Ophthalmol. 150, 25â32, (2025). 35.Ergun Yildiz, M. et al. Retinal electrophysiological responses to dopaminergic therapy in treatment naĂŻve parkinsonâs disease. Documenta Ophthalmol. 1â13, (2026). 36.Constable, P. A., Lee, I. O., Marmolejo-Ramos, F., Skuse, D. H. & Thompson, D. A. The photopic negative response in autism spectrum disorder. Clin. Exp. Optom. 104, 841â847, (2021). 37.Friedel, E. B. et al. Electroretinography in adults with high-functioning autism spectrum disorder. Autism Res. 15, 2026â2037, (2022). 38. Constable, P. A., Lim, J. K. & Thompson, D. A. Retinal electrophysiology in central nervous system disorders. a review of human and mouse studies. Front. Neurosci. 17, 1215097, (2023). 39.Soto Linan, V., Hebert, M. & Levesque, M. Enhancing the diagnostic potential of electroretinography in parkinsonâs disease: A review of protocol and cohort criteria. J. Park. Dis. 15, 694â709, (2025). 40.Hebert, M. et al. The electroretinogram may differentiate schizophrenia from bipolar disorder. Biol. psychiatry 87, 263â270, (2020). 41.Gauthier, M., Polosa, A., Lina, J.-M. & Lachapelle, P. Hilbert transform analysis of the mouse scotopic electroretinogram reveals two distinct bursts of oscillatory potentials with progressively dimmer flashes. Documenta Ophthalmol. 150, 1â15, (2025). 42.Hamilton, R., Bees, M., Chaplin, C. A. & McCulloch, D. L. The luminanceâresponse function of the human photopic electroretinogram: A mathematical model. Vis. research 47, 2968â2972, (2007). 43.McCulloch, D. L. et al. Iscev extended protocol for the stimulusâresponse series for light-adapted full-field erg. Documenta Ophthalmol. 138, 205â215, (2019). 44. Mahroo, O. & Lamb, T. D. Recovery of the human photopic electroretinogram after bleaching exposures: estimation of pigment regeneration kinetics. The J. physiology 554, 417â437, (2004). 45.Tzekov, R., Kremers, J. & Safari, S. The area under the waveform as an alternative measure of the photopic negative response. Documenta Ophthalmol. 1â12, (2026). 46.Prencipe, M., Perossini, T., Brancoli, G. & Perossini, M. The photopic negative response (phnr): measurement approaches and utility in glaucoma. Int. Ophthalmol. 40, 3565â3576, (2020). 47.Silverstein, S. M., Fradkin, S. I. & Demmin, D. L. Schizophrenia and the retina: towards a 2020 perspective. Schizophr. research 219, 84â94, (2020). 48.Constable, P. A., Gaigg, S. B., Bowler, D. M., JĂ€gle, H. & Thompson, D. A. Full-field electroretinogram in autism spectrum disorder. Documenta Ophthalmol. 132, 83â99, (2016). 49. Ritvo, E. R. et al. Electroretinograms in autism: a pilot study of b-wave amplitudes. Am J Psychiatry 145, 229â32, (1988). 50.Hosseinzadeh, N. et al. Optical coherence tomography measurements of retina and choroid in autism spectrum disorder: a systematic review and meta-analysis. Ophthalmic Res. 68, 531â554, (2025). 51.Kulyabin, M. et al. Enhancing electroretinogram classification with multi-wavelet analysis and visual transformer. Sensors 23, 8727, (2023). 52.Kulyabin, M., Zhdanov, A., Dolganov, A. & Maier, A. Optimal combination of mother wavelet and ai model for precise classification of pediatric electroretinogram signals. Sensors 23, 5813, (2023). 11/12 53.Peredo, R., HĂ©bert, M. & MĂ©rette, C. Developing a clinical decision tool based on electroretinogram to monitor the risk of severe mental illness. BMC psychiatry 22, 718, (2022). 54. Constable, P. Erg dataset, 10.25451/flinders.14747349, (2021). 55. FernĂĄndez, I., Cuadrado-Asensio, R., Larriba, Y., Rueda, C. & Coco-MartĂn, R. M. A comprehensive dataset of pattern electroretinograms for ocular electrophysiology research. Sci. Data 11, 1013, (2024). 56. Wang, J., Wang, Y., Guan, W. & Zhao, Y.-E. Full-field electroretinogram recorded with skin electrodes in 6-to 12-year-old children. Documenta Ophthalmol. 147, 179â188, (2023). 57.Chan, S. S.-h. et al. Reference ranges of light-adapted full-field electroretinogram and associated factors in a large cohort of healthy school-aged children and adolescents. Documenta Ophthalmol. 151, 3â17, (2025). 58.Baker, R. A. et al. Iscev standard full-field erg reference limits from 407 healthy subjects, derived from transference and validation of reference data between electrode types and centres. Documenta Ophthalmol. 150, 47â64, (2025). 59. Kulyabin, M. et al. Synthetic electroretinogram signals for enhancing classification of autism spectrum disorder, 10.21227/ npv7-8063, (2024). 60. Constable, P. A. et al. Light-adapted electroretinogram differences in autism spectrum disorder. J. Autism Dev. Disord. 50, 2874â2885, (2020). 61.Gotham, K., Pickles, A. & Lord, C. Standardizing ados scores for a measure of severity in autism spectrum disorders. J. autism developmental disorders 39, 693â705, (2009). 62. Schopler, E., Reichler, R. J. & Renner, B. R. The childhood autism rating scale (CARS) (Western Psychological Services Los Angeles, CA, 2010). 63. APA. Diagnostic and statistical manual of mental disorders V (American Psychiatric Association Publishing, 2022). 64. APA. Diagnostic and statistical manual of mental disorders 4th edition. Am Psychiatr. Assoc 21, 591â643, (2013). 65. Lord, C. et al. ADOS-2: Autism diagnostic observation schedule (Western Psychological Services, 2012). 66. Arias-Alvarez, M., Sopena-Pinilla, M., Rodriguez-Mena, D. & Pinilla, I. The mydriasis-free handheld erg device and its utility in clinical practice: A review. Biomedicines 14, 384, (2026). 67.You, J. Y. et al. Comparing the reteval portable erg device with more traditional tabletop erg systems in normal subjects and selected retinopathies. Documenta Ophthalmol. 146, 137â150, (2023). 68.Porciatti, V. & Ventura, L. M. Normative data for a user-friendly paradigm for pattern electroretinogram recording. Ophthalmology 111, 161â168, (2004). Author contributions statement Data collection, P.C, I.L, L.L. and D.T. conceptualization, P.C and M.K. software, M.K, A.M. and A.Z. writing-original draft preparation, P.C., and M.K writingâreview and editing, P.C., M.K., D.T and A.M supervision, P.C, D.T. and A.M. All authors have read and agreed to the published version of the manuscript. Competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We thank Quentin Davis and Joshua Santosa of LKC Technologies for the custom RETeval protocols. The authors acknowledge Professor Emeritus Edward Ritvo for his inspiration and support for investigating the electroretinogram in ASD. We would also thank Professor David Skuse for his support in establishing the study in the UK. 12/12