My experience at APELA: personalisation for people with ALS

The Organization

Associação Portuguesa de Esclerose Lateral Amiotrófica (APELA) is Portugal's major ALS association, providing comprehensive care and support for patients living with amyotrophic lateral sclerosis and their families. I’m taking this opportunity to write about my visits to APELA in the context of my internship with Unbabel Labs' research team. The main product of the team was a non-invasive neural interface combining electromyography (EMG) sensors with generative AI to enable communication through bioelectrical signals. APELA served as both a collaborative partner and a testing ground. On their own, APELA represented a model of care where physical therapists not only treated motor impairments; but also architected entire digital ecosystems personalized to each patient's specific needs, abilities, and patterns of movement.

The goal

My role centered on building an AI persona platform that allowed caretakers, both family members and APELA's physical therapists, to input detailed information about individual ALS patients. This data would feed into personalized language models, enabling the system to predict and generate communication more effectively tailored to each person. Through APELA, I worked with specific patients, their families and took input from physical therapists.

What I witnessed there fundamentally reshaped my understanding of what accessibility actually means in practice.

Each patient used different assistive technology with varying degrees of effectiveness. Some relied on GRID 3—a comprehensive AAC (Augmentative and Alternative Communication) software system that functions almost as an entirely different operating system, featuring symbol-based vocabularies, text-to-speech with message banking (allowing people to record phrases in their own voice while they still can), and sophisticated word prediction. Others used eye-tracking devices like Tobii Dynavox's TD I-Series or Eyegaze Edge systems, which require careful calibration and can be physically exhausting, particularly outdoors or in challenging lighting conditions. The variety alone was striking, but what truly astonished me was the depth of personalization embedded in every interface - custom keyboards for individual patients created manually by the therapists.

These keyboards that predicted the next word or letter, unique to each patient’s communication patterns, vocabulary, and cognitive model. If one patient used a particular app significantly more than others, the therapist would enlarge that icon and position it precisely where the patient's range of motion (whether in the upper body, neck, or eye movement) could access it most efficiently. Even the colors chosen for each app icon relied on studies about visual processing and cognitive load. Everything was engineered to reduce response times, to minimize the gap between thought and expression.

The GRID system itself became a personalized universe. For patients preparing for the inevitable loss of speech, APELA had purchased specific laptops pre-loaded with AAC software, but this introduced an unexpected barrier: older patients, many unfamiliar with computers, had to simultaneously learn to navigate an unfamiliar operating system while also adapting to a radically new mode of communication. The therapists worked to bridge this gap, but it highlighted how accessibility isn't just about providing technology. It's about understanding the full context of someone's life, their technical literacy, their cognitive patterns, and their emotional state.

I met with several patients, always with a physical therapist present, and spoke with their families. We were conducting user research in its truest form: observing what technologies people currently use, understanding their frustrations and workarounds, to create interface elements together. I returned a couple times, bringing back refined versions of our prototype and watching patients interact with the dashboard I developed.

Personalization as the future of access

Before APELA, I understood accessibility primarily as accommodation—as making spaces, objects, and systems usable by the greatest number of people. I thought in terms of universal design, of broad inclusivity. What I learned at APELA was that true accessibility, at least in the context of degenerative conditions like ALS, isn't universal at all. It's radically particular.

The physical therapists at APELA went far beyond the medical. They were information architects, and interface designers. In a technological age where an entirely separate world exists online, facilitating patient access to that world is a fundamental component of care, very much in line with access to the internet being a human right. This expanded definition of a physical therapist's role revealed something essential: disability and accessibility culture recognizes that participation in digital life isn't supplementary to quality of life. In fact, it is quality of life.

This experience crystallized a tension I'd been feeling in my own work. Capitalistic forces drive us toward creating single personas, standardised solutions that can scale and appeal to mass markets. This is what I was taught in business classes but in this context - and I’m sure, many others - personalization isn’t a premium feature. It was the baseline requirement for participation.

This raises urgent questions for the future I'm now grappling within my work on the future of personal devices. How do we make this kind of personalisation economically sustainable, creating incentives for it to be pursued from the get go? Not only that, but also considering that many patients, especially in Portugal may not have access to this organisation (which is one of a kind in my country) or to people to set up our software for them in their laptop, like we were doing? How do we truly build for everyone?

I suspect the answer has to do with someone actually intentionally building for this group of people using AI to create interfaces walking this very thin line of anything is possible without going for typical AI lingo that only specialists or the younger generation understands. Paired with changing the education on what prototyping digital services means.

The true lesson is recognising that every person is an edge case, rightfully so, and that technology's job isn't to smooth away that complexity but to meet it with equal sophistication.