Traditional therapies often lack the necessary diversity and adaptability to effectively engage individuals with Neurodevelopmental Disorders (NDDs), which encompass conditions characterized by cognitive and communication difficulties. The emergence of Generative AI and advancements in Large Language Models (LLMs) such as GPT-4 offer promising new avenues for customized and interactive therapeutic approaches. LLMs have an exceptional ability to comprehend and generate text in a manner similar to human communication, providing a flexible foundation for creating personalized interventions through prompt engineering. GPT-4, in particular, marks a significant milestone in the Human-Computer Interaction (HCI) domain, standing out as one of the most advanced LLMs for supporting the development of interactive and tailored therapeutic methods. We introduce ChatCare, the Conversational Helper for Assisted Therapy, Caregiver Assistance, and Recreational Education. ChatCare harnesses recent technological advancements to aid individuals with NDDs by enhancing communication, educational out- comes, and emotional comprehension. A key feature of ChatCare is its ability to utilize LLMs to generate real-time, customized social stories tailored to the user’s needs while contextualizing them within a dynamic conversation. This technology can be further ex- panded thanks to a standardized and formal design, making ChatCare highly modular and easily extendable with additional exercises as necessary. Additionally, an alternative interface to the original has been developed to distinguish between the conversational and exercise components, offering more delivery options for the system. In our research, we validated the effectiveness of LLM technology for individuals with NDDs by organizing a test day at an educational center in Milan. After the initial release of ChatCare, we held a focus group with seven experts in the field of NDDs to gather professional feedback on the application.
Le terapie tradizionali spesso mancano della necessaria diversità e adattabilità per coinvolgere efficacemente individui con Disturbi del Neurosviluppo (NDD), che comprendono condizioni caratterizzate da difficoltà cognitive e comunicative. L’emergere dell’Intelligenza Artificiale Generativa e i progressi nei Modelli Linguistici di Grandi Dimensioni (LLM) come GPT-4 offrono nuove promettenti strade per approcci terapeutici personalizzati e interattivi. Gli LLM possiedono un’eccezionale capacità di comprendere e generare testo in modo simile alla comunicazione umana, fornendo una base flessibile per creare interventi personalizzati attraverso l’ingegneria dei prompt. GPT-4, in particolare, segna un significativo traguardo nel dominio dell’Interazione Uomo-Macchina (HCI), distinguendosi come uno dei LLM più avanzati per supportare lo sviluppo di metodi terapeutici interattivi e su misura. Presentiamo ChatCare, (Conversational Helper for Assisted Therapy, Caregiver Assistance, and Recreational Education). ChatCare sfrutta i recenti progressi tecnologici per aiutare gli individui con NDD migliorando la comunicazione, i risultati educativi e la comprensione emotiva. Una caratteristica chiave di ChatCare è la sua capacità di utilizzare gli LLM per generare in tempo reale storie sociali personalizzate adattate alle esigenze dell’utente, contestualizzandole all’interno di una conversazione dinamica. Questa tecnologia può essere ulteriormente ampliata grazie a un design standardizzato e formale, rendendo ChatCare altamente modulare e facilmente estendibile con attività aggiuntive secondo necessità. Inoltre, è stata sviluppata un’interfaccia alternativa a quella originale per distinguere tra la componente conversazionale e quella delle attività, offrendo più opzioni di erogazione per il sistema. Nella nostra ricerca, abbiamo validato l’efficacia della tecnologia LLM per gli individui con NDD organizzando una giornata di test presso un centro educativo a Milano. Dopo il rilascio iniziale di ChatCare, abbiamo organizzato un focus group con sette esperti nel campo degli NDD per raccogliere feedback professionali sull’applicazione.
ChatCare: a generative AI-based tool for creating social stories for individuals with neurodevelopmental disorders
CALEFFI, GIOVANNI
2023/2024
Abstract
Traditional therapies often lack the necessary diversity and adaptability to effectively engage individuals with Neurodevelopmental Disorders (NDDs), which encompass conditions characterized by cognitive and communication difficulties. The emergence of Generative AI and advancements in Large Language Models (LLMs) such as GPT-4 offer promising new avenues for customized and interactive therapeutic approaches. LLMs have an exceptional ability to comprehend and generate text in a manner similar to human communication, providing a flexible foundation for creating personalized interventions through prompt engineering. GPT-4, in particular, marks a significant milestone in the Human-Computer Interaction (HCI) domain, standing out as one of the most advanced LLMs for supporting the development of interactive and tailored therapeutic methods. We introduce ChatCare, the Conversational Helper for Assisted Therapy, Caregiver Assistance, and Recreational Education. ChatCare harnesses recent technological advancements to aid individuals with NDDs by enhancing communication, educational out- comes, and emotional comprehension. A key feature of ChatCare is its ability to utilize LLMs to generate real-time, customized social stories tailored to the user’s needs while contextualizing them within a dynamic conversation. This technology can be further ex- panded thanks to a standardized and formal design, making ChatCare highly modular and easily extendable with additional exercises as necessary. Additionally, an alternative interface to the original has been developed to distinguish between the conversational and exercise components, offering more delivery options for the system. In our research, we validated the effectiveness of LLM technology for individuals with NDDs by organizing a test day at an educational center in Milan. After the initial release of ChatCare, we held a focus group with seven experts in the field of NDDs to gather professional feedback on the application.File | Dimensione | Formato | |
---|---|---|---|
2024_07_Caleffi_Executive_Summary_01.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
2.72 MB
Formato
Adobe PDF
|
2.72 MB | Adobe PDF | Visualizza/Apri |
2024_07_Caleffi_Tesi_01.pdf
accessibile in internet per tutti
Descrizione: Tesi
Dimensione
63.64 MB
Formato
Adobe PDF
|
63.64 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/223161