noemi gozzi

We’re thrilled to see our advanced ML models and EMG hardware — that transform neural signals controlling muscles at the wrist into commands that seamlessly drive computer interactions — appearing in the latest edition of @Nature. Read the story: nature.com/articles/s4158… Find more details on this work and the models on @github: github.com/facebookresear…

🚨New Paper Out in @LancetDigitalH🚨 "Clinical trials for implantable neural prostheses: understanding the ethical and technical requirements" Collaborative work with brilliant co-authors @giacomo_valle and @SRaspopovic. 🔎 Why this matters: Neuroprosthetics research is rapidly moving from proof-of-concept studies to real-world clinical applications, often powered by #AI. However, existing clinical trial frameworks may not be fully equipped to address the complexity of implantable neural prostheses due to insufficient validation mechanisms for AI algorithms embedded in these devices and inadequate assessment of the neural prosthesis' impact on users' subjective experience (e.g. sense of agency). 💡 Some of our key takeaways: 1) Clinical trials must be amended to assess not just safety and efficacy, but also embodiment, user experience, agency and other phenomenological impacts. 2) To do so we need better quantitative-qualitative assessments to comprehensively study the impact of neural prostheses on users' subjective experience. 3) Privacy and data security must be integral to safety evaluations during trials, including assessments of mental privacy risks. 4) Mitigating algorithmic bias requires incorporating explainable AI (XAI) into neuroprostheses and using more population-representative datasets. 5) Trials should address the potential misalignments in user agency caused by AI-based decision-making. 6) Companies must ensure robust financial and operational safeguards to prevent forced explantation if the device provider fails. We advocate for patient-centred research ethics that go beyond standard protocols by incorporating privacy impact assessments, robust liability frameworks, and long-term participant support. Read the full paper here: thelancet.com/journals/landi…