DETECTAA-AI: Artificial Intelligence in the presumptive diagnosis of developmental disorders in children

Editorial note: this article documents an educational project and its results on the data used. It does not establish clinical validation or authorisation to diagnose or treat patients. Proposed healthcare applications are retained as the team’s goals, not demonstrated capabilities.

Quito. 2021

We use Artificial Intelligence to help us make the presumptive diagnosis of disorders in school-aged children.

Saturdays.AI is a global initiative, whose principle is to promote scenarios for the democratization of Artificial Intelligence learning for everyone and ubiquitously. Democratizing means facilitating access to all citizens who wish to achieve pertinent, relevant and quality training, at any educational or professional level. For this reason, the research and development team, made up of: {Andrea Mariana Escobar, Danny Aguirre, Luis Chamba Eras, Marco Chiluiza, Paúl Quezada}, decided to participate in the Third Edition of Saturdays AI Quito, which in an unprecedented, ubiquitous and flexible way, was developed virtually.

In the first session, brainstorming was developed, with the objective of identifying the base line of research, on which the project would be developed, especially that it has a social impact and related to the objectives-goals of the Agenda 2030.

Originally the topic “Chatbot for the management of emotions in autistic children” was proposed, obtaining the first problem tree (Fig. 1), then the literature search strategy was launched to define the scope of the proposal, 27 scientific articles linked to that baseline were found (see Table 1).

Figure 1. Initial problem tree.

The scientific literature allowed us to know and understand what has been done and what can be done on issues with autism, with this it was concluded that the topic is very broad and with a lot of future work for projects linked to the computer science part with a social purpose. Furthermore, it was identified that there is no free access data set that serves as a starting point for the topic raised.

Another key point was to search in groups related to the topic of autism, both on social networks and on the Web, with this it was observed that it is a very delicate and complex topic, from the point of view of those who live with autism, or those of us who do not. Possibly it is an issue that has not had the visibility and democratization that allows defining policies to support and educate all of us who interact with people with autism, whether directly or indirectly. With this, it was necessary to go to professionals or specialists in the field, to clear up many doubts that arose from the preliminary exploration, and thus see the viability of the proposal.

Along the way new clues emerged, a data set was found on Kaggle (https://www.kaggle.com/gpiosenka/autistic-children-data-set-traintestvalidate), related to autism, which has been used to build some models that allow, through computer vision, to predict from a photograph whether or not a child has autism. With this, the perspective of the project changed, from moving from emotions (without a data set) to facial recognition (with a data set) in the same area of ​​autism.

To continue to know the professional opinion on the proposal, two interviews were carried out, the first with the specialist Amparito Morales, to whom we presented our new idea, that through technology we could help improve diagnoses in the area of autism, initially, there was resistance in the use of technology, but that was good, because it allowed us as a team to convince the professional of its real usefulness in a scenario such as in large schools or schools, where the work of the few specialists (Department of Student Counseling (DECE)) can be supported by a tool that supports automation tasks, in this case, recognizing which of the children through a photograph could have priority attention in the early detection of autism.

From the first interview emerged the second, with the renowned researcher Catalina Lopez, a pioneer in Ecuador for its sensory-perceptive approach to identify autism profiles according to the idiosyncrasy of a country.

Currently, she is finishing a screening tool aimed at children and adolescents from 4 to 17 years old (features to alert clinical diagnosis). In addition, during the interview, Catalina López validated the idea of the project, adding new ideas linked to technologies, and that have emerged from her research, such as, for example, virtual reality to apply the screening tools, automation of the screening tool considering data protection, privacy, anonymity, confidentiality, code of ethics under principles worldwide, informed consent, among others.

Finally, the researcher proposed that a chatbot, through interaction, whether by voice or text, would allow the identification of behavior patterns and the theme of emotions. This interview established the work or future lines that derive from the project, focusing on the issue of easy recognition and a screening tool (Fig. 3), as the limit for the final proposal of the DETECTAA-AI project, with which we worked on Saturdays AI.

Figure 2. Interview with Catalina López, Specialist in Human Communication Disturbances at the Simón Bolívar Andean University.

Figure 3. Brainstorming the initial model of the DETECTAA-AI project.

Context

Developmental disorders, technically known as neurodevelopmental disorders, are neurologically based disorders that can affect the acquisition, retention, or application of specific skills or sets of information. They consist of alterations in attention, memory, perception, language, problem solving or social interaction. These disorders can be mild and easily addressed with behavioral and educational interventions or more serious, so that affected children require particular educational support. Among neurodevelopmental disorders we have: attention deficit/hyperactivity disorder, autism spectrum disorders, learning difficulties, such as dyslexia and deficiencies in other academic areas, intellectual disability, Rett syndrome.

“Autism is a complex neurological disorder that usually lasts a lifetime. It is part of a group of disorders known as autism spectrum disorders (ASD). Currently, 1 in 68 individuals and 1 in 42 male children are diagnosed with autism, making it more common than pediatric cancer, diabetes, and AIDS combined. It occurs in any racial, ethnic and social group, and is four times more common in boys than in girls. Autism impairs a person's ability to communicate and relate to others. Also, it is associated with repetitive routines and behaviors, such as obsessively arranging objects or following very specific routines. Symptoms can range from mild to very severe” [1].

Autism in Ecuador

According to the specialist Catalina López, there are the following advances:

• Pilot study for the concurrent validation of the Social Communication Questionnaire (SCQ), as a screening instrument for the detection of the autism spectrum in a sample of the Ecuadorian population aged 0 to 12 years (2016). Technical Report.
• Autism in Ecuador: a Social Group Waiting for Care (2017). Article in Magazine.
• Comparative study of conceptions about autism, from the perspective of neurosciences and neurodiversity (2019). Technical Report.

Worldwide, it is estimated that 1% may have ASD, According to the World Health Organization, in 2018 1,521 were reported in Ecuador, and approximately 13.75% have erroneous diagnoses.

What is the problem?

The staff who work in the student counseling departments of the educational units (DECE) must carry out evaluations to determine which students may present behavioral problems. Due to the large number of students assigned to each professional in these departments, the evaluation process consumes the greatest amount of time available to this staff, leaving very few resources to deepen the diagnosis and support of children who truly present developmental disorders. Fig. 4 shows the problem tree, which was obtained after brainstorming, reading the literature and after the interviews.

Figure 4. Tree of problems related to the DETECTAA-AI project.

How do we plan to solve it?

A Web application will be developed consisting of two components (Fig. 3).

The first component will help predict which students may or may not have ASD based on a photographic image (passport size) using computer vision. The traits determined depend on the available databases. In the first phase, the database available on Kaggle will be used (https://www.kaggle.com/gpiosenka/autistic-children-data-set-traintestvalidate) for facial detection of ASD, considering defining a system training process that allows detecting new behavioral factors as additional image bases become available.

Technically, screening corresponds to the application of a test or procedure to “asymptomatic” people, with the aim of separating them into two groups; those who have a condition that could benefit from early intervention; and those who don't.

The second component will perform a screening, using the MCHAT test, and that is the basis for implementing natural language processing (question and answer chatbot) in the future.

How is the project linked to sustainable development objectives?

It is linked to two objectives:

First, with the Health and well-being (SDG 3), goal: strengthen the capacity of all countries, particularly developing countries, in early warning, risk reduction and risk management for national and global health.

Second, with the Reducing inequalities (SDG 10), goal: progress in reducing inequality, both within and between countries, has been uneven. The opinion of developing countries still needs to be given more weight in the decision-making forums of international economic and financial institutions. Furthermore, while remittances can be a means of survival for the families and communities of international migrant workers in their home countries, the high cost of transferring money continues to reduce the benefits.

What is the hypothesis of the project?

The use of Artificial Intelligence will allow the creation of a prototype that supports the presumptive diagnosis of developmental disorders in school-age children.

What is the target population?

• Children from 0 to 12 years
• Fathers, mothers, caregivers
• Educators
• DECE specialists
• Researchers

What does the scientific literature tell us about projects related to facial recognition?

The scientific literature that supports our project is summarized in the Table 2.

What is computer vision?

It is a field of Artificial Intelligence focused on computers being able to extract information from images, offering solutions to real-world problems (Fig. 5).

Figure 5. Facial recognition can help improve diagnoses, right photo, child without ASD, child on the left, child with ASD.

What areas of knowledge are linked?

• Health Sciences (Mental Health).
• Computer Science (Artificial Intelligence, Computer Vision).

Methodology

The methodology used was Design Thinking; Fig. 6 shows a summary of each of the stages developed.

Figure 6. Description of each of the stages of the methodology according to the DETECTAA-AI project.

In Fig. 7, there is a working canvas provided by https://www.analogolab.co/, to implement the principles of the Design Thinking methodology. In this web link, the complete design of the project is observed.

Figure 7. General idea of the project, Mapping of actors linked to the project, defining clients or stakeholders in the project, future beneficiaries, Declaration of the idea, Positive factors, opportunities, problems and solutions.

Results

Architecture

The project architecture is divided into a Frontend application and a Backend application (see Fig. 8). The Frontend, developed with Flask (Python Framework), contains all the interfaces with which the end user interacts. This, in turn, is connected through an endpoint to the Backend. In the Backend there is an API, developed with Flask, that contains a Deep Learning model trained with libraries of TensorFlow and a set of images obtained from Kaggle. The Frontend also interacts with a model trained on Teachable Machine (a Google platform to train machine learning models quickly and easily).

Figure 8. Proposed architecture for DETECTAA-AI.

Web links to the APIs and the DETECTAA-AI application:

• Teachable Machine: https://teachablemachine.withgoogle.com/models/5O0q09n5c/
• Tensorflow Model API: https://reconocimientoteaapi.herokuapp.com/
• FrontEnd: https://detectai.herokuapp.com/

DETECTAA-AI Workflow

The results obtained for the first case (child with ASD) are quite favorable, since both the models and the questionnaire give a high percentage of detection of ASD in the person evaluated, as shown in Fig. 9.

Figure 9. Workflow, case 1.

The results of the second case (girl without ASD) present acceptable percentages in the diagnosis of ASD. As Fig. 10 shows, the results obtained were: Teachable Machine: 100%, TensorFlow: 85.28% and M-Chat: Low Risk.

Figure 10. Workflow, case 2.

In the third case (child without ASD) the results obtained from the models and M-chat reflect different results, since the machine learning models return accurate diagnoses regarding the test performed, however, the M-chat returns a high risk of having a diagnosis of ASD, as shown in Fig. 11.

Figure 11. Workflow, case 3.

Conclusions

With the development of the DETECTAA-AI project, the following conclusions were reached:

• It is possible to detect signs of ASD in people through the use of artificial intelligence models.
• For a model to have a higher reliability rate, a greater number of training images and better processing of that information is necessary.
• Artificial intelligence algorithms serve as support for health professionals, but not as a replacement.
• A link between the academy, the state, companies, unions, and societies is necessary so that these initiatives can be launched according to the Ecuadorian context.
• Combining scientific research with professional processes allows us to build scalable prototypes over time.
• The DETECTAA-AI prototype should be used for academic and research purposes, as an example of proof of concept, and not to be offered as a final diagnostic tool, since a team of professionals is needed to contribute to the detection of ASD.

Líneas futuras

• Implement the NLP screening tool, of a question and answer type, using the questionnaire proposed by Dr. Catalina López in the Ecuadorian context, considering privacy, data protection, among others.
• Obtain your own database of images in the context of Ecuador, to test the DETECTAA-AI prototype.
• It is advisable to increase a third ASD detection tool by NLP, which allows detecting the presence of ASD by analyzing patterns in the voice of the person who needs to be diagnosed.
• Raise awareness among the population that technology can be a very important support in the context of Health.

DETECTAA-AI project resources

• Teachable Machine Page: https://teachablemachine.withgoogle.com/
• Dataset of training and testing images: https://www.kaggle.com/gpiosenka/autistic-children-data-set-traintestvalidate
• GitHub repository: https://github.com/detectaai
• API Link: recognitionteaapi.herokuapp.com
• Final Application Link: https://detectai.herokuapp.com/

References

[1] https://www.uasb.edu.ec/reconocimiento-a-la-directora-del-area-de-salud-catalina-lopez-id1550289/

Project presentation: DemoDay

Open video or resource

More artificial intelligence!

The mission of Saturdays.ai is to make artificial intelligence more accessible (#ai4all) through intensive courses and programs where projects for good are carried out (#ai4good).

Explore our AI master’s programme at https://saturdays.ai/master-ia-online/

If you want to learn more about artificial intelligence join our community at community.saturdays.ai or visit us on our website www.saturdays.ai Join us! Saturdays.AI

Public domain.

• Machine Learning
• Ecuador
• ADHD
• Artificial Intelligence
• Ai Saturdays

WRITTEN BY

Luis Chamba-Eras

Professor and researcher at the National University of Loja. Research in Artificial Intelligence in Education.

Saturdays.AI

Saturdays.AI is an impact-focused organization on a mission to empower diverse individuals to learn Artificial Intelligence in a collaborative and project-based way, beyond the conventional path of traditional education.