How ABA Therapists View Machine Learning: Insights from a National Survey

Introduction

Applied Behavior Analysis (ABA) is a widely used therapy for individuals with Autism Spectrum Disorder (ASD). A core component of ABA is consistent data collection, which helps therapists track behavior changes over time and adjust interventions accordingly.

However, data collection in ABA therapy is often time-consuming and prone to human error. Therapists must manually record behaviors, which can be tedious and affect accuracy. As a potential solution, **machine learning (ML)**—a type of artificial intelligence (AI)—is being explored to streamline data collection and analysis.

A recent study, Perceptions of Machine Learning among Therapists Practicing Applied Behavior Analysis: A National Survey (Doan et al., 2024), surveyed ABA therapists across the U.S. to understand their views on ML in data collection. Here’s what they found.

Common Data Collection Methods in ABA

Data collection in ABA therapy helps therapists assess progress and modify treatment plans. Current methods include:

1. Continuous recording – Recording every instance of a behavior in real time.
2. Interval recording – Observing behavior at set intervals (e.g., partial-interval, whole-interval, or momentary time sampling).
3. Paper and pencil methods – Manually jotting down behavioral observations.
4. Digital data collection apps – Using software to log and analyze behavioral data.
5. Video recording – Capturing sessions on video for later analysis.
6. Wearable technology – Devices that track movement, heart rate, or other physiological data.
7. Smart speaker or voice command logs – Using voice-activated devices to document behavioral data hands-free.

While digital tools and automation have improved efficiency, challenges remain. Traditional methods can be inconsistent, and therapists may struggle with capturing subtle behavior changes. This is where machine learning can step in.

Understanding Machine Learning in ABA

Machine learning is a type of AI that can process large amounts of data, recognize patterns, and make predictions. In healthcare, ML has been used to:

1. Predict disease progression – Identifying patterns in patient data to foresee health outcomes.
2. Analyze movement patterns in autism – Using ML to recognize behavioral traits characteristic of ASD.
3. Detect suicide risks on social media – AI analyzing online behavior to identify at-risk individuals.
4. Recommend personalized therapy plans – Customizing interventions based on individual response patterns.
5. Automate documentation and reporting – Reducing administrative workload for therapists.
6. Improve early intervention assessments – Enhancing early detection and diagnosis of ASD.

In ABA therapy, ML could help therapists analyze behavioral trends more accurately, reduce manual data collection errors, and allow for real-time adjustments to treatment plans.

How the Study Was Conducted

Doan et al. (2024) surveyed ABA practitioners across the United States, including:

• Registered Behavior Technicians (RBTs)
• Board Certified Assistant Behavior Analysts (BCaBAs)
• Board Certified Behavior Analysts (BCBAs and BCBA-Ds)
• Board Certified Autism Technicians (BCATs)

The survey gathered data on:

1. Demographic details of participants.
2. Their current data collection methods.
3. Their familiarity with and perceptions of ML in ABA therapy.

Researchers then analyzed responses to determine how therapists feel about integrating ML into their work.

Key Findings: Therapists’ Views on Machine Learning

1. Awareness of Machine Learning in ABA

• 39.52% of therapists surveyed were already familiar with ML.
• Those familiar with ML were more comfortable with the idea of using it in ABA.
• Many therapists who lacked ML knowledge were skeptical about its accuracy and reliability.

2. Comfort with ML for Data Collection

• Overall, therapists were neutral or slightly supportive of using ML instead of manual data collection.
• Some therapists expressed concerns about ML’s ability to recognize subtle nuances in behavior.
• A major barrier was the perceived “black box” nature of AI, meaning therapists found it difficult to understand how ML arrives at its conclusions.

3. Perceived Benefits and Concerns

Therapists saw ML as beneficial in the following ways:

• Reducing manual errors – Automating data entry could improve accuracy.
• Increasing efficiency – Less time spent recording data means more time focusing on therapy.
• Better long-term trend detection – ML could uncover behavior patterns that may not be obvious through traditional data tracking.

However, concerns included:

• Potential loss of clinician expertise – Therapists worried that over-reliance on AI could undervalue their judgment.
• Privacy and ethical concerns – Issues around data security and consent were raised.
• Limited insight into ML mechanisms – Without clear explanations of how ML models work, therapists were hesitant to trust them.

What This Means for ABA Therapy

Machine learning has the potential to transform ABA therapy by improving the accuracy and efficiency of data collection. However, this study highlights that many therapists remain uncertain about how ML works and whether it can be trusted.

For ML to gain acceptance in ABA:

• Education is key – Training programs should introduce ABA therapists to ML concepts and real-world applications.
• Transparency matters – ML tools need to explain how they analyze data, so therapists feel confident in their outputs.
• Collaboration is essential – Developers should work alongside ABA experts to ensure ML tools align with the field’s needs.

Final Thoughts

As technology continues to evolve, ABA therapists will likely see more AI-driven tools entering their field. While this study shows some hesitation toward ML, it also suggests that with proper education and transparency, therapists may be more open to integrating AI into their practice.

If you’d like to explore the full study, check it out here: Perceptions of Machine Learning among Therapists Practicing Applied Behavior Analysis: A National Survey by Doan et al. (2024).

What do you think—can machine learning help improve ABA data collection, or is it too soon to rely on AI in therapy? Let us know in the comments!