Elia Eiroa-Lledo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Elia Eiroa-Lledo has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Erik Linstead
Fowler School of Engineering, Chapman University, Orange, CA, United States
4 shared publications Abigail Atchison
Chapman University,Fowler School of Engineering,Orange,CA
1 shared publicationAdrienne Bergh
Chapman University,Fowler School of Engineering,Orange,CA
1 shared publication- Alicia Brewer Curran
ECHO Autism Communities, University of Missouri School of Medicine, Columbia, MO, United States
1 shared publication - Carmela Salomon
Cognoa Inc, 5000 Campus Drive, Newport Beach, CA, 92663, United States, 1 6507852624
1 shared publication - Elizabeth Stevens
Fowler School of Engineering, Chapman University, Orange, CA 92866, USA
1 shared publication
Latest Publications
- Open Access
- 21 October 2025
- Published by JMIR Publications Inc. in JMIR Formative Research
- Vol. 9, e80733
Integration of an Artificial Intelligence–Based Autism Diagnostic Device into the ECHO Autism Primary Care Workflow: Prospective Observational Study
Background: Pediatric specialist shortages and rapidly rising autism prevalence rates have compelled primary care clinicians to consider playing a greater role in the autism diagnostic process. The ECHO Autism: Early Diagnosis Program (EDx) prepares clinicians to screen, evaluate, differentiate, diagnose, and provide longitudinal care for children with autism in primary care settings. Canvas Dx is a prescription-only Software as a Medical Device designed to support clinical diagnosis or rule out of autism, including in primary care settings. It is authorized by the Food and Drug Administration for use, in conjunction with clinical judgment, in 18‐ to 72-month-olds with indicators of developmental delay. Objective: This study aims to assess the feasibility and impact of integrating the device into the ECHO Autism: EDx workflow. Time from the first clinical question of developmental delay to autism diagnosis is the primary endpoint. Secondary endpoints explore clinician and caregiver experience of device use. Methods: Children aged 18 to 72 months with concern for developmental delay indicated by either a caregiver or health professionals were eligible to participate in this prospective observational study. Experienced ECHO Autism: EDx clinicians were recruited to evaluate the inclusion of the device as part of their diagnostic evaluations. Outcome data were collected via a combination of electronic questionnaires, standard clinical care record reviews, and analysis of device outputs. Institutional review board approval was provided by the University of Missouri-Columbia (project number 2075722). Results: Eighty children and 7 clinicians completed the study. On average, time from clinical concern at study enrollment to final autism diagnosis was 39.22 days, compared to 180‐ to 264-day waits at adjacent specialist referral centers. The vast majority (93%, 50/54) of caregivers reported being satisfied with the ECHO Autism: EDx plus device evaluation their child received and endorsed that they would recommend it to others and that they felt comfortable using the device. The device produced determinate autism predictions or rule-outs for 52.5% of participants, and in all cases, these were consistent with the final clinical determination. Participating clinicians reported that device use was feasible and reduced several challenges associated with their previous diagnostic process; however, they noted it did not obviate the need for additional structured observation in every case. Conclusions: The ECHO Autism: EDx plus device workflow offers considerable time savings compared to specialty center referral and was strongly endorsed by caregiver participants. Embedding the device into the ECHO Autism: EDx workflow was feasible and helped streamline several workflow efficiencies. Clinicians still utilized their training and application and interpretation of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, criteria when formulating the diagnosis for indeterminate cases.
- 21 October 2025
- Published by JMIR Publications Inc. in JMIR Formative Research
- Vol. 9, e80733
Open AccessIntegration of an Artificial Intelligence–Based Autism Diagnostic Device into the ECHO Autism Primary Care Workflow: Prospective Observational Study
Background: Pediatric specialist shortages and rapidly rising autism prevalence rates have compelled primary care clinicians to consider playing a greater role in the autism diagnostic process. The ECHO Autism: Early Diagnosis Program (EDx) prepares clinicians to screen, evaluate, differentiate, diagnose, and provide longitudinal care for children with autism in primary care settings. Canvas Dx is a prescription-only Software as a Medical Device designed to support clinical diagnosis or rule out of autism, including in primary care settings. It is authorized by the Food and Drug Administration for use, in conjunction with clinical judgment, in 18‐ to 72-month-olds with indicators of developmental delay. Objective: This study aims to assess the feasibility and impact of integrating the device into the ECHO Autism: EDx workflow. Time from the first clinical question of developmental delay to autism diagnosis is the primary endpoint. Secondary endpoints explore clinician and caregiver experience of device use. Methods: Children aged 18 to 72 months with concern for developmental delay indicated by either a caregiver or health professionals were eligible to participate in this prospective observational study. Experienced ECHO Autism: EDx clinicians were recruited to evaluate the inclusion of the device as part of their diagnostic evaluations. Outcome data were collected via a combination of electronic questionnaires, standard clinical care record reviews, and analysis of device outputs. Institutional review board approval was provided by the University of Missouri-Columbia (project number 2075722). Results: Eighty children and 7 clinicians completed the study. On average, time from clinical concern at study enrollment to final autism diagnosis was 39.22 days, compared to 180‐ to 264-day waits at adjacent specialist referral centers. The vast majority (93%, 50/54) of caregivers reported being satisfied with the ECHO Autism: EDx plus device evaluation their child received and endorsed that they would recommend it to others and that they felt comfortable using the device. The device produced determinate autism predictions or rule-outs for 52.5% of participants, and in all cases, these were consistent with the final clinical determination. Participating clinicians reported that device use was feasible and reduced several challenges associated with their previous diagnostic process; however, they noted it did not obviate the need for additional structured observation in every case. Conclusions: The ECHO Autism: EDx plus device workflow offers considerable time savings compared to specialty center referral and was strongly endorsed by caregiver participants. Embedding the device into the ECHO Autism: EDx workflow was feasible and helped streamline several workflow efficiencies. Clinicians still utilized their training and application and interpretation of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, criteria when formulating the diagnosis for indeterminate cases.
- Open Access
- 28 February 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (5) , 3150
Large-Scale Identification and Analysis of Factors Impacting Simple Bug Resolution Times in Open Source Software Repositories
One of the most prominent issues the ever-growing open-source software community faces is the abundance of buggy code. Well-established version control systems and repository hosting services such as GitHub and Maven provide a checks-and-balances structure to minimize the amount of buggy code introduced. Although these platforms are effective in mitigating the problem, it still remains. To further the efforts toward a more effective and quicker response to bugs, we must understand the factors that affect the time it takes to fix one. We apply a custom traversal algorithm to commits made for open source repositories to determine when “simple stupid bugs” were first introduced to projects and explore the factors that drive the time it takes to fix them. Using the commit history from the main development branch, we are able to identify the commit that first introduced 13 different types of simple stupid bugs in 617 of the top Java projects on GitHub. Leveraging a statistical survival model and other non-parametric statistical tests, we found that there were two main categories of categorical variables that affect a bug’s life; Time Factors and Author Factors. We find that bugs are fixed quicker if they are introduced and resolved by the same developer. Further, we discuss how the day of the week and time of day a buggy code was written and fixed affects its resolution time. These findings will provide vital insight to help the open-source community mitigate the abundance of code and can be used in future research to aid in bug-finding programs.
- 28 February 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (5) , 3150
Open AccessLarge-Scale Identification and Analysis of Factors Impacting Simple Bug Resolution Times in Open Source Software Repositories
One of the most prominent issues the ever-growing open-source software community faces is the abundance of buggy code. Well-established version control systems and repository hosting services such as GitHub and Maven provide a checks-and-balances structure to minimize the amount of buggy code introduced. Although these platforms are effective in mitigating the problem, it still remains. To further the efforts toward a more effective and quicker response to bugs, we must understand the factors that affect the time it takes to fix one. We apply a custom traversal algorithm to commits made for open source repositories to determine when “simple stupid bugs” were first introduced to projects and explore the factors that drive the time it takes to fix them. Using the commit history from the main development branch, we are able to identify the commit that first introduced 13 different types of simple stupid bugs in 617 of the top Java projects on GitHub. Leveraging a statistical survival model and other non-parametric statistical tests, we found that there were two main categories of categorical variables that affect a bug’s life; Time Factors and Author Factors. We find that bugs are fixed quicker if they are introduced and resolved by the same developer. Further, we discuss how the day of the week and time of day a buggy code was written and fixed affects its resolution time. These findings will provide vital insight to help the open-source community mitigate the abundance of code and can be used in future research to aid in bug-finding programs.
- Open Access
- 15 March 2021
- Published by MDPI AG in Electronics
- Vol. 10 (6) , 689
On-Device Deep Learning Inference for System-on-Chip (SoC) Architectures
As machine learning becomes ubiquitous, the need to deploy models on real-time, embedded systems will become increasingly critical. This is especially true for deep learning solutions, whose large models pose interesting challenges for target architectures at the “edge” that are resource-constrained. The realization of machine learning, and deep learning, is being driven by the availability of specialized hardware, such as system-on-chip solutions, which provide some alleviation of constraints. Equally important, however, are the operating systems that run on this hardware, and specifically the ability to leverage commercial real-time operating systems which, unlike general purpose operating systems such as Linux, can provide the low-latency, deterministic execution required for embedded, and potentially safety-critical, applications at the edge. Despite this, studies considering the integration of real-time operating systems, specialized hardware, and machine learning/deep learning algorithms remain limited. In particular, better mechanisms for real-time scheduling in the context of machine learning applications will prove to be critical as these technologies move to the edge. In order to address some of these challenges, we present a resource management framework designed to provide a dynamic on-device approach to the allocation and scheduling of limited resources in a real-time processing environment. These types of mechanisms are necessary to support the deterministic behavior required by the control components contained in the edge nodes. To validate the effectiveness of our approach, we applied rigorous schedulability analysis to a large set of randomly generated simulated task sets and then verified the most time critical applications, such as the control tasks which maintained low-latency deterministic behavior even during off-nominal conditions. The practicality of our scheduling framework was demonstrated by integrating it into a commercial real-time operating system (VxWorks) then running a typical deep learning image processing application to perform simple object detection. The results indicate that our proposed resource management framework can be leveraged to facilitate integration of machine learning algorithms with real-time operating systems and embedded platforms, including widely-used, industry-standard real-time operating systems.
- 15 March 2021
- Published by MDPI AG in Electronics
- Vol. 10 (6) , 689
Open AccessOn-Device Deep Learning Inference for System-on-Chip (SoC) Architectures
As machine learning becomes ubiquitous, the need to deploy models on real-time, embedded systems will become increasingly critical. This is especially true for deep learning solutions, whose large models pose interesting challenges for target architectures at the “edge” that are resource-constrained. The realization of machine learning, and deep learning, is being driven by the availability of specialized hardware, such as system-on-chip solutions, which provide some alleviation of constraints. Equally important, however, are the operating systems that run on this hardware, and specifically the ability to leverage commercial real-time operating systems which, unlike general purpose operating systems such as Linux, can provide the low-latency, deterministic execution required for embedded, and potentially safety-critical, applications at the edge. Despite this, studies considering the integration of real-time operating systems, specialized hardware, and machine learning/deep learning algorithms remain limited. In particular, better mechanisms for real-time scheduling in the context of machine learning applications will prove to be critical as these technologies move to the edge. In order to address some of these challenges, we present a resource management framework designed to provide a dynamic on-device approach to the allocation and scheduling of limited resources in a real-time processing environment. These types of mechanisms are necessary to support the deterministic behavior required by the control components contained in the edge nodes. To validate the effectiveness of our approach, we applied rigorous schedulability analysis to a large set of randomly generated simulated task sets and then verified the most time critical applications, such as the control tasks which maintained low-latency deterministic behavior even during off-nominal conditions. The practicality of our scheduling framework was demonstrated by integrating it into a commercial real-time operating system (VxWorks) then running a typical deep learning image processing application to perform simple object detection. The results indicate that our proposed resource management framework can be leveraged to facilitate integration of machine learning algorithms with real-time operating systems and embedded platforms, including widely-used, industry-standard real-time operating systems.
- 1 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1276-1280
A Curated Set of Labeled Code Tutorial Images for Deep Learning
As more work is conducted in mining software repositories at a large scale, aggregating data from multiple modalities beyond textual representations will prove beneficial. Online video content related to software engineering offers another, thus far largely untapped, venue for source code. In this paper, we present a novel dataset comprised of 111,229 frames extracted from a sample of 100 technology-related YouTube tutorials, taught in the Java and Python programming languages. These videos span a diverse range of Integrated Development Environments (IDEs) and modes of information presentation, lending heterogeneity of image features to our dataset. Hand-labeled for the presence, or lack thereof, of typeset, partially visible typeset, or handwritten code, the dataset provides a uniquely consistent and accurate data source for supervised machine learning ventures in extracting and processing code from images.
- 1 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1276-1280
A Curated Set of Labeled Code Tutorial Images for Deep Learning
As more work is conducted in mining software repositories at a large scale, aggregating data from multiple modalities beyond textual representations will prove beneficial. Online video content related to software engineering offers another, thus far largely untapped, venue for source code. In this paper, we present a novel dataset comprised of 111,229 frames extracted from a sample of 100 technology-related YouTube tutorials, taught in the Java and Python programming languages. These videos span a diverse range of Integrated Development Environments (IDEs) and modes of information presentation, lending heterogeneity of image features to our dataset. Hand-labeled for the presence, or lack thereof, of typeset, partially visible typeset, or handwritten code, the dataset provides a uniquely consistent and accurate data source for supervised machine learning ventures in extracting and processing code from images.