{ "id": "AIBD-0005", "title": "First direct decoding of full words from the cortex of a paralyzed, anarthric man (UCSF, BRAVO trial)", "summary": "128-electrode ECoG and deep-learning decoder let a man mute since 2003 produce sentences at 15.2 wpm from a 50-word vocabulary", "date_observed": "2019-04-01", "date_basis": "unassessed", "date_precision": "unassessed", "benefit_status": null, "source_collection": null, "source_entry": null, "account": "At the University of California, San Francisco, a 36-year-old man received a 128-electrode subdural electrocorticography array over sensorimotor cortex in the BRAVO trial (NCT03698149). He had anarthria and severe spastic quadriparesis from an extensive pontine stroke at age 20 and is known in press by the pseudonym Pancho. Deep-learning models detected his attempted speech and classified words from a 50-word vocabulary, with a natural-language model and Viterbi decoder assembling sentences. He produced sentences at a median 15.2 words per minute with a median word error rate of 25.6%. He previously typed about five words per minute with a head-controlled pointer (Moses et al., NEJM, 15 July 2021). The paper dates implantation to February 2019 and the start of data collection to April 2019. There were 50 sessions over 81 weeks at the participant's residence or a nearby office. The session on which the reported sentence decoding was achieved is not dated. So the start of data collection (1 April 2019) is used as the observation date.", "category": "benefit created", "unexpected": false, "purpose": "Built to decode attempted speech into words and sentences for a person with anarthria. Restoring word-level communication was the intended benefit and is what occurred.", "system": "UCSF speech neuroprosthesis (BRAVO trial)", "model_version": "128-electrode high-density ECoG array (16-by-8 lattice, 4 mm spacing); deep-learning speech-detection and word-classification models with natural-language model and Viterbi decoder; 50-word vocabulary", "developer": "UCSF Chang Lab", "deployer": "University of California, San Francisco", "location": "San Francisco, United States", "domain": "Healthcare / assistive communication", "segment": "Post-stroke anarthria", "benefit_type": "Improved Accessibility", "taxonomy_note": null, "registry_ids": {}, "ratings": [ { "group": "personal", "depth": 2, "reason": "Material benefit to one identifiable person: first decoding of his own full words, three times faster than his pointer typing. But limited to a 50-word vocabulary in research sessions." } ], "ai_role": { "status": "assessed", "statement": "Deep-learning models detected attempted speech in 98% of attempts and classified words. A language model with Viterbi decoding assembled sentences. The output depended entirely on these models. Funding included a research contract under Facebook's Sponsored Academic Research Agreement. The paper states that no study hardware or data were transferred to any sponsor and that author disclosure forms are available at NEJM.org." }, "evidence": [ { "kind": "peer-reviewed study", "citation": "Moses DA, Metzger SL, Liu JR, et al. (2021). Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria. N Engl J Med.", "title": "Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/", "date": "2021-07-15", "written_by": "makers", "quote": "median rate of 15.2 words per minute, with a median word error rate of 25.6%" }, { "kind": "news report", "citation": "Advisory Board (2021). For the first time since 2003, this paralyzed man can 'speak'. Advisory Board Daily Briefing.", "title": "For the first time since 2003, this paralyzed man can 'speak'", "url": "https://www.advisory.com/daily-briefing/2021/07/16/paralyzed-man", "date": "2021-07-16", "written_by": "independent", "quote": "To our knowledge, this is the first successful demonstration of direct decoding of full words from the brain activity of someone who is paralyzed and cannot speak." }, { "kind": "other", "citation": "Reis e Sousa C, Poirier E (2021). Scientists Detect Brain Signals to Help a Paralyzed Man Speak. Harvard Science in the News.", "title": "Scientists Detect Brain Signals to Help a Paralyzed Man Speak", "url": "https://sites.harvard.edu/sitn/2021/09/22/scientists-detect-brain-signals-to-help-a-paralyzed-man-speak", "date": "2021-09-22", "written_by": "independent", "quote": "In this way, the electrodes allowed Pancho to communicate about 15 words per minute with near 50% accuracy." } ], "replication": { "produced_by": [ "Preserved speech-motor cortical pathways more than 15 years after loss of speech", "High-density ECoG array over sensorimotor cortex", "Deep-learning models trained on 22 hours of attempted-speech data plus language-model decoding" ], "must_hold": [ "Participant can attempt speech and retains intact cognition", "Implant remains stable with adequate signal quality", "Accumulated training data maintained over time" ], "echoes": [ { "date": "2023-08-23", "setting": "UCSF, a different participant (47-year-old woman, brainstem stroke 2005): 253-channel ECoG, 78 words per minute, 1,024-word vocabulary, voice and avatar", "model_version": "253-channel ECoG; bidirectional RNN decoders", "outcome": "recurred", "source": "Metzger SL, et al. (2023). A high-performance neuroprosthesis for speech decoding and avatar control. Nature. https://pmc.ncbi.nlm.nih.gov/articles/PMC10826467/" }, { "date": "2024-05-20", "setting": "UCSF, same participant (the full text at PMC11554235 describes a 36-year-old Spanish-English bilingual with anarthria and severe spastic quadriparesis after a bilateral pontine stroke, a 128-electrode ECoG array and the BRAVO trial, and refers to initial work with this participant): bilingual (Spanish-English) sentence decoding without specifying the target language", "model_version": "ECoG with deep-learning and bilingual natural-language models", "outcome": "recurred", "source": "Silva AB, et al. (2024). A bilingual speech neuroprosthesis driven by cortical articulatory representations shared between languages. Nature Biomedical Engineering. doi:10.1038/s41551-024-01207-5" } ] }, "paired_incident": { "database": "AI Incident Database", "identifier": "732", "title": "Whisper Speech-to-Text AI Reportedly Found to Create Violent Hallucinations", "url": "https://incidentdatabase.ai/cite/732", "severity": null, "group": "personal", "basis": "Both decode speech for people with impaired speech; Whisper fabricated words and violent phrases when transcribing impaired speech." }, "headline": "UCSF neuroprosthesis lets Pancho produce sentences at 15.2 wpm", "metric": "sentences at a median 15.2 words per minute with a median word error rate of 25.6%", "comparator": null, "later": null, "value": null, "developer_country": "United States", "deployer_country": "United States", "unrated_because": null, "brief_notes": null, "brief": null, "revisions": [ { "date": "2026-10-20", "change": "Record drafted from the cited sources.", "source": null, "editor": "AIBD" }, { "date": "2026-10-20", "change": "Paired with AI Incident Database incident 732 (medium confidence); severity not given by the source.", "source": "https://incidentdatabase.ai/cite/732", "editor": "AIBD" }, { "date": "2026-10-20", "change": "Moved the date observed from the publication date to April 2019, when the paper says data collection began after the February 2019 implant, replaced the unsupported “no competing interests” claim with the paper's sponsor and disclosure-form statements, added a taxonomy note on why a research-session demonstration is filed as a benefit created or restored, and verified the Silva 2024 echo as the same participant against the PMC full text.", "source": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/; https://pmc.ncbi.nlm.nih.gov/articles/PMC11554235/", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Country of the developer and the deployer added from the organization's own site or a registry.", "source": "https://changlab.ucsf.edu/; https://en.wikipedia.org/wiki/University_of_California,_Berkeley", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Prose pass: a rating reason, account, purpose, the system's part rewritten in shorter sentences with abbreviations expanded; facts, figures and quotations unchanged.", "source": null, "editor": "AIBD" } ], "confidence": "reported", "shade": "medium", "claim_status": "Independently reported", "claim_basis": "Sources: one independent news report, one other independent source and the makers' peer-reviewed study. The cited evidence includes no independent study or report and no finding by a regulator or court.", "profile": { "case_id": "AIBD-0005", "catalog_version": "1.0.0", "source_fingerprint": "fb4b93baaa4cdf0674a02044c4ef2d2b8d707541300e829f2c347ec9fc3a4d5a", "collection_refs": [ { "collection_id": "aibd", "version": "snapshot-8bf7cfec9cb185aa" }, { "collection_id": "aibd", "version": "2026-10-25" } ], "source_ids": [ "AIBD-0005-source-0001", "AIBD-0005-source-0002", "AIBD-0005-source-0003" ], "families": { "strategic assessments": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "record_ids": [] }, "evaluations": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "record_ids": [] }, "monetary observations": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "record_ids": [] }, "institutional decisions": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "record_ids": [] } }, "fields": { "actors[]": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "developer": "UCSF Chang Lab", "deployer": "University of California, San Francisco" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "developer", "deployer" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "actors[].ownership_control": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "ai_contribution": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "ai_role": { "status": "assessed", "statement": "Deep-learning models detected attempted speech in 98% of attempts and classified words. A language model with Viterbi decoding assembled sentences. The output depended entirely on these models. Funding included a research contract under Facebook's Sponsored Academic Research Agreement. The paper states that no study hardware or data were transferred to any sponsor and that author disclosure forms are available at NEJM.org." } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ai_role" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "ai_relationship": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "category": "benefit created", "ai_role": { "status": "assessed", "statement": "Deep-learning models detected attempted speech in 98% of attempts and classified words. A language model with Viterbi decoding assembled sentences. The output depended entirely on these models. Funding included a research contract under Facebook's Sponsored Academic Research Agreement. The paper states that no study hardware or data were transferred to any sponsor and that author disclosure forms are available at NEJM.org." } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ai_role", "category" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "assessments[].alternatives": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "assessments[].assessor_date": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "assessments[].depth": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "ratings": [ { "group": "personal", "depth": 2, "reason": "Material benefit to one identifiable person: first decoding of his own full words, three times faster than his pointer typing. But limited to a 50-word vocabulary in research sessions." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ratings" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "assessments[].display_confidence": { "state": "value_present", "value": { "kind": "structured", "data": { "confidence": "medium", "evidence_word": "reported", "claim_status": "Independently reported", "basis": "Sources: one independent news report, one other independent source and the makers' peer-reviewed study. The cited evidence includes no independent study or report and no finding by a regulator or court.", "interpretation": "Existing display rule preserved; the richer outcome-specific confidence question still needs review." } }, "provenance": { "kind": "curatorial", "basis": "Derived from the existing case display properties; not a new evidentiary judgment.", "record_fields": [], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "assessments[].rationale": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "ratings": [ { "group": "personal", "depth": 2, "reason": "Material benefit to one identifiable person: first decoding of his own full words, three times faster than his pointer typing. But limited to a 50-word vocabulary in research sessions." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ratings" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "assessments[].rubric_version": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "assessments[].status_basis": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "assessments[].target": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "ratings": [ { "group": "personal", "depth": 2, "reason": "Material benefit to one identifiable person: first decoding of his own full words, three times faster than his pointer typing. But limited to a 50-word vocabulary in research sessions." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ratings" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "availability": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "brief": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "case_kind": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "case_links[]": { "state": "value_present", "value": { "kind": "structured", "data": { "answer": "Both decode speech for people with impaired speech; Whisper fabricated words and violent phrases when transcribing impaired speech.", "links": [ { "database": "AI Incident Database", "identifier": "732", "title": "Whisper Speech-to-Text AI Reportedly Found to Create Violent Hallucinations", "url": "https://incidentdatabase.ai/cite/732", "group": "personal", "basis": "Both decode speech for people with impaired speech; Whisper fabricated words and violent phrases when transcribing impaired speech.", "relation": "comparable_harm" } ] } }, "provenance": { "kind": "legacy_record", "basis": "Existing paired_incident classification; no new source review.", "record_fields": [ "paired_incident" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "claims[].evidence_links": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "claims[].evidence_strength": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "claims[].missingness": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "claims[].proposition": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "cross_border_flows[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "decisions[].action_outcome": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "institutional decisions" }, "decisions[].grounds": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "institutional decisions" }, "decisions[].institution_task": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "institutional decisions" }, "decisions[].success_criterion": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "institutional decisions" }, "deployment_stage": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "economics[].amount": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].costs[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].counterfactual": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].currency_price_basis": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].gross_net": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].measure_type": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].period_unit": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].realization": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].recipient_payer": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].return_payback": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "economics[].valuation_method": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "monetary observations" }, "evaluations[].ai_uplift": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].comparator": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].design": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].effect": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].exposure": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].id": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].measurement_method": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].metric": { "state": "not_assessed", "provenance": { "kind": "legacy_record", "basis": "Mapped from the existing case description without a new external source review.", "record_fields": [ "metric" ], "source_refs": [] }, "review_state": "not_assessed", "coverage": "partial", "family": "evaluations", "display": { "answer": "sentences at a median 15.2 words per minute with a median word error rate of 25.6%", "editor_note": "" } }, "evaluations[].reproduction_materials": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].setting_sample": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "evaluations[].uncertainty_limitations": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "evaluations" }, "external_references[]": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "paired_incident": { "database": "AI Incident Database", "identifier": "732", "title": "Whisper Speech-to-Text AI Reportedly Found to Create Violent Hallucinations", "url": "https://incidentdatabase.ai/cite/732", "group": "personal", "basis": "Both decode speech for people with impaired speech; Whisper fabricated words and violent phrases when transcribing impaired speech." } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "paired_incident" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "follow_up[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "human_oversight": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "id": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "id": "AIBD-0005" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "id" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "infrastructure_dependencies[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].beneficiaries": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].beneficiary_count": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].benefit_type": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "benefit_type": "Improved Accessibility" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "benefit_type" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "outcomes[].claim": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "account": "At the University of California, San Francisco, a 36-year-old man received a 128-electrode subdural electrocorticography array over sensorimotor cortex in the BRAVO trial (NCT03698149). He had anarthria and severe spastic quadriparesis from an extensive pontine stroke at age 20 and is known in press by the pseudonym Pancho. Deep-learning models detected his attempted speech and classified words from a 50-word vocabulary, with a natural-language model and Viterbi decoder assembling sentences. He produced sentences at a median 15.2 words per minute with a median word error rate of 25.6%. He previously typed about five words per minute with a head-controlled pointer (Moses et al., NEJM, 15 July 2021). The paper dates implantation to February 2019 and the start of data collection to April 2019. There were 50 sessions over 81 weeks at the participant's residence or a nearby office. The session on which the reported sentence decoding was achieved is not dated. So the start of data collection (1 April 2019) is used as the observation date.", "benefit_type": "Improved Accessibility" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "account", "benefit_type" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "outcomes[].distribution": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].duration": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].id": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].stakeholder_group": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "ratings": [ { "group": "personal", "depth": 2, "reason": "Material benefit to one identifiable person: first decoding of his own full words, three times faster than his pointer typing. But limited to a 50-word vocabulary in research sessions." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "ratings" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "outcomes[].status": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].tradeoffs": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "outcomes[].unexpected": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "unexpected": false, "purpose": "Built to decode attempted speech into words and sentences for a person with anarthria. Restoring word-level communication was the intended benefit and is what occurred." } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "unexpected", "purpose" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "places[]": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "location": "San Francisco, United States", "developer_country": "United States", "deployer_country": "United States" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "location", "developer_country", "deployer_country" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "purpose": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "purpose": "Built to decode attempted speech into words and sentences for a person with anarthria. Restoring word-level communication was the intended benefit and is what occurred." } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "purpose" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "record_dates": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "date_observed": "2019-04-01", "revisions": [ { "date": "2026-10-20", "change": "Record drafted from the cited sources.", "editor": "AIBD" }, { "date": "2026-10-20", "change": "Paired with AI Incident Database incident 732 (medium confidence); severity not given by the source.", "source": "https://incidentdatabase.ai/cite/732", "editor": "AIBD" }, { "date": "2026-10-20", "change": "Moved the date observed from the publication date to April 2019, when the paper says data collection began after the February 2019 implant, replaced the unsupported “no competing interests” claim with the paper's sponsor and disclosure-form statements, added a taxonomy note on why a research-session demonstration is filed as a benefit created or restored, and verified the Silva 2024 echo as the same participant against the PMC full text.", "source": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/; https://pmc.ncbi.nlm.nih.gov/articles/PMC11554235/", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Country of the developer and the deployer added from the organization's own site or a registry.", "source": "https://changlab.ucsf.edu/; https://en.wikipedia.org/wiki/University_of_California,_Berkeley", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Prose pass: a rating reason, account, purpose, the system's part rewritten in shorter sentences with abbreviations expanded; facts, figures and quotations unchanged.", "editor": "AIBD" } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "date_observed", "revisions" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "replication.attempts[]": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "replication": { "produced_by": [ "Preserved speech-motor cortical pathways more than 15 years after loss of speech", "High-density ECoG array over sensorimotor cortex", "Deep-learning models trained on 22 hours of attempted-speech data plus language-model decoding" ], "must_hold": [ "Participant can attempt speech and retains intact cognition", "Implant remains stable with adequate signal quality", "Accumulated training data maintained over time" ], "echoes": [ { "date": "2023-08-23", "setting": "UCSF, a different participant (47-year-old woman, brainstem stroke 2005): 253-channel ECoG, 78 words per minute, 1,024-word vocabulary, voice and avatar", "model_version": "253-channel ECoG; bidirectional RNN decoders", "outcome": "recurred", "source": "Metzger SL, et al. (2023). A high-performance neuroprosthesis for speech decoding and avatar control. Nature. https://pmc.ncbi.nlm.nih.gov/articles/PMC10826467/" }, { "date": "2024-05-20", "setting": "UCSF, same participant (the full text at PMC11554235 describes a 36-year-old Spanish-English bilingual with anarthria and severe spastic quadriparesis after a bilateral pontine stroke, a 128-electrode ECoG array and the BRAVO trial, and refers to initial work with this participant): bilingual (Spanish-English) sentence decoding without specifying the target language", "model_version": "ECoG with deep-learning and bilingual natural-language models", "outcome": "recurred", "source": "Silva AB, et al. (2024). A bilingual speech neuroprosthesis driven by cortical articulatory representations shared between languages. Nature Biomedical Engineering. doi:10.1038/s41551-024-01207-5" } ] } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "replication" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "replication.conditions": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "replication": { "produced_by": [ "Preserved speech-motor cortical pathways more than 15 years after loss of speech", "High-density ECoG array over sensorimotor cortex", "Deep-learning models trained on 22 hours of attempted-speech data plus language-model decoding" ], "must_hold": [ "Participant can attempt speech and retains intact cognition", "Implant remains stable with adequate signal quality", "Accumulated training data maintained over time" ], "echoes": [ { "date": "2023-08-23", "setting": "UCSF, a different participant (47-year-old woman, brainstem stroke 2005): 253-channel ECoG, 78 words per minute, 1,024-word vocabulary, voice and avatar", "model_version": "253-channel ECoG; bidirectional RNN decoders", "outcome": "recurred", "source": "Metzger SL, et al. (2023). A high-performance neuroprosthesis for speech decoding and avatar control. Nature. https://pmc.ncbi.nlm.nih.gov/articles/PMC10826467/" }, { "date": "2024-05-20", "setting": "UCSF, same participant (the full text at PMC11554235 describes a 36-year-old Spanish-English bilingual with anarthria and severe spastic quadriparesis after a bilateral pontine stroke, a 128-electrode ECoG array and the BRAVO trial, and refers to initial work with this participant): bilingual (Spanish-English) sentence decoding without specifying the target language", "model_version": "ECoG with deep-learning and bilingual natural-language models", "outcome": "recurred", "source": "Silva AB, et al. (2024). A bilingual speech neuroprosthesis driven by cortical articulatory representations shared between languages. Nature Biomedical Engineering. doi:10.1038/s41551-024-01207-5" } ] } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "replication" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "revision_history[]": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "revisions": [ { "date": "2026-10-20", "change": "Record drafted from the cited sources.", "editor": "AIBD" }, { "date": "2026-10-20", "change": "Paired with AI Incident Database incident 732 (medium confidence); severity not given by the source.", "source": "https://incidentdatabase.ai/cite/732", "editor": "AIBD" }, { "date": "2026-10-20", "change": "Moved the date observed from the publication date to April 2019, when the paper says data collection began after the February 2019 implant, replaced the unsupported “no competing interests” claim with the paper's sponsor and disclosure-form statements, added a taxonomy note on why a research-session demonstration is filed as a benefit created or restored, and verified the Silva 2024 echo as the same participant against the PMC full text.", "source": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/; https://pmc.ncbi.nlm.nih.gov/articles/PMC11554235/", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Country of the developer and the deployer added from the organization's own site or a registry.", "source": "https://changlab.ucsf.edu/; https://en.wikipedia.org/wiki/University_of_California,_Berkeley", "editor": "AIBD" }, { "date": "2026-10-21", "change": "Prose pass: a rating reason, account, purpose, the system's part rewritten in shorter sentences with abbreviations expanded; facts, figures and quotations unchanged.", "editor": "AIBD" } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "revisions" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "sector": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "domain": "Healthcare / assistive communication", "segment": "Post-stroke anarthria" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "domain", "segment" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "sources[].access_version": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "sources[].citation_locator": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "evidence": [ { "kind": "peer-reviewed study", "citation": "Moses DA, Metzger SL, Liu JR, et al. (2021). Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria. N Engl J Med.", "title": "Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/", "date": "2021-07-15", "written_by": "makers", "quote": "median rate of 15.2 words per minute, with a median word error rate of 25.6%" }, { "kind": "news report", "citation": "Advisory Board (2021). For the first time since 2003, this paralyzed man can 'speak'. Advisory Board Daily Briefing.", "title": "For the first time since 2003, this paralyzed man can 'speak'", "url": "https://www.advisory.com/daily-briefing/2021/07/16/paralyzed-man", "date": "2021-07-16", "written_by": "independent", "quote": "To our knowledge, this is the first successful demonstration of direct decoding of full words from the brain activity of someone who is paralyzed and cannot speak." }, { "kind": "other", "citation": "Reis e Sousa C, Poirier E (2021). Scientists Detect Brain Signals to Help a Paralyzed Man Speak. Harvard Science in the News.", "title": "Scientists Detect Brain Signals to Help a Paralyzed Man Speak", "url": "https://sites.harvard.edu/sitn/2021/09/22/scientists-detect-brain-signals-to-help-a-paralyzed-man-speak", "date": "2021-09-22", "written_by": "independent", "quote": "In this way, the electrodes allowed Pancho to communicate about 15 words per minute with near 50% accuracy." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "evidence" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "sources[].id": { "state": "value_present", "value": { "kind": "identifiers", "identifiers": [ "AIBD-0005-source-0001", "AIBD-0005-source-0002", "AIBD-0005-source-0003" ], "role": "Existing evidence items, in their original order" }, "provenance": { "kind": "curatorial", "basis": "Stable within the fingerprinted record version; no source content or source review has been added.", "record_fields": [], "source_refs": [] }, "review_state": "record_mapped", "coverage": "complete" }, "sources[].language": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "sources[].relationship": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "evidence": [ { "kind": "peer-reviewed study", "citation": "Moses DA, Metzger SL, Liu JR, et al. (2021). Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria. N Engl J Med.", "title": "Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8972947/", "date": "2021-07-15", "written_by": "makers", "quote": "median rate of 15.2 words per minute, with a median word error rate of 25.6%" }, { "kind": "news report", "citation": "Advisory Board (2021). For the first time since 2003, this paralyzed man can 'speak'. Advisory Board Daily Briefing.", "title": "For the first time since 2003, this paralyzed man can 'speak'", "url": "https://www.advisory.com/daily-briefing/2021/07/16/paralyzed-man", "date": "2021-07-16", "written_by": "independent", "quote": "To our knowledge, this is the first successful demonstration of direct decoding of full words from the brain activity of someone who is paralyzed and cannot speak." }, { "kind": "other", "citation": "Reis e Sousa C, Poirier E (2021). Scientists Detect Brain Signals to Help a Paralyzed Man Speak. Harvard Science in the News.", "title": "Scientists Detect Brain Signals to Help a Paralyzed Man Speak", "url": "https://sites.harvard.edu/sitn/2021/09/22/scientists-detect-brain-signals-to-help-a-paralyzed-man-speak", "date": "2021-09-22", "written_by": "independent", "quote": "In this way, the electrodes allowed Pancho to communicate about 15 words per minute with near 50% accuracy." } ] } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "evidence" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "sources[].upstream_sources": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "strategic_assessments[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial", "family": "strategic assessments" }, "summary": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "summary": "128-electrode ECoG and deep-learning decoder let a man mute since 2003 produce sentences at 15.2 wpm from a 50-word vocabulary", "account": "At the University of California, San Francisco, a 36-year-old man received a 128-electrode subdural electrocorticography array over sensorimotor cortex in the BRAVO trial (NCT03698149). He had anarthria and severe spastic quadriparesis from an extensive pontine stroke at age 20 and is known in press by the pseudonym Pancho. Deep-learning models detected his attempted speech and classified words from a 50-word vocabulary, with a natural-language model and Viterbi decoder assembling sentences. He produced sentences at a median 15.2 words per minute with a median word error rate of 25.6%. He previously typed about five words per minute with a head-controlled pointer (Moses et al., NEJM, 15 July 2021). The paper dates implantation to February 2019 and the start of data collection to April 2019. There were 50 sessions over 81 weeks at the participant's residence or a nearby office. The session on which the reported sentence decoding was achieved is not dated. So the start of data collection (1 April 2019) is used as the observation date." } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "summary", "account" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].autonomy_permissions": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "systems[].component_role": { "state": "value_present", "value": { "kind": "text", "text": "The models detected attempted speech, classified words, and assembled sentences." }, "provenance": { "kind": "legacy_record", "basis": "Restated the mechanical task already described in the existing system and AI-role fields. No new source review is claimed.", "record_fields": [ "system", "ai_role" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].deployment_configuration": { "state": "value_present", "value": { "kind": "text", "text": "The system used a 128-electrode array with 4 mm spacing and a vocabulary of 50 words." }, "provenance": { "kind": "legacy_record", "basis": "Separated the model, software, hardware, or operation already described in the existing record. No new source review is claimed.", "record_fields": [ "model_version", "system", "ai_role" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].inputs_outputs": { "state": "value_present", "value": { "kind": "text", "text": "The system detected attempted speech in cortical recordings and classified words from a 50-word vocabulary." }, "provenance": { "kind": "legacy_record", "basis": "Separated the model, software, hardware, or operation already described in the existing record. No new source review is claimed.", "record_fields": [ "system", "ai_role", "replication", "account" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].model_name": { "state": "value_present", "value": { "kind": "text", "text": "Deep-learning models detect speech and classify words. A language model and a Viterbi decoder assemble the output." }, "provenance": { "kind": "legacy_record", "basis": "Separated the model, software, hardware, or operation already described in the existing record. No new source review is claimed.", "record_fields": [ "model_version", "system", "ai_role" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].model_version": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "model_version": "128-electrode high-density ECoG array (16-by-8 lattice, 4 mm spacing); deep-learning speech-detection and word-classification models with natural-language model and Viterbi decoder; 50-word vocabulary" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "model_version" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial", "display": { "answer": "Exact release not established.", "editor_note": "" } }, "systems[].product_name": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "system": "UCSF speech neuroprosthesis (BRAVO trial)" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "system" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "systems[].provider": { "state": "value_present", "value": { "kind": "structured", "data": { "application_developer": "UCSF Chang Lab" } }, "provenance": { "kind": "legacy_record", "basis": "Mapped from the existing case description without a new external source review.", "record_fields": [ "developer" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "taxonomy_crosswalks[]": { "state": "not_assessed", "review_state": "not_assessed", "coverage": "partial" }, "title": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "title": "First direct decoding of full words from the cortex of a paralyzed, anarthric man (UCSF, BRAVO trial)", "headline": "UCSF neuroprosthesis lets Pancho produce sentences at 15.2 wpm" } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "title", "headline" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" }, "version": { "state": "value_present", "value": { "kind": "identifiers", "identifiers": [ "fb4b93baaa4cdf0674a02044c4ef2d2b8d707541300e829f2c347ec9fc3a4d5a" ], "role": "SHA-256 of normalized original case data" }, "provenance": { "kind": "curatorial", "basis": "Deterministic content version assigned by the local migration.", "record_fields": [], "source_refs": [] }, "review_state": "record_mapped", "coverage": "complete" }, "workflow": { "state": "value_present", "value": { "kind": "record_fields", "fields": { "account": "At the University of California, San Francisco, a 36-year-old man received a 128-electrode subdural electrocorticography array over sensorimotor cortex in the BRAVO trial (NCT03698149). He had anarthria and severe spastic quadriparesis from an extensive pontine stroke at age 20 and is known in press by the pseudonym Pancho. Deep-learning models detected his attempted speech and classified words from a 50-word vocabulary, with a natural-language model and Viterbi decoder assembling sentences. He produced sentences at a median 15.2 words per minute with a median word error rate of 25.6%. He previously typed about five words per minute with a head-controlled pointer (Moses et al., NEJM, 15 July 2021). The paper dates implantation to February 2019 and the start of data collection to April 2019. There were 50 sessions over 81 weeks at the participant's residence or a nearby office. The session on which the reported sentence decoding was achieved is not dated. So the start of data collection (1 April 2019) is used as the observation date.", "replication": { "produced_by": [ "Preserved speech-motor cortical pathways more than 15 years after loss of speech", "High-density ECoG array over sensorimotor cortex", "Deep-learning models trained on 22 hours of attempted-speech data plus language-model decoding" ], "must_hold": [ "Participant can attempt speech and retains intact cognition", "Implant remains stable with adequate signal quality", "Accumulated training data maintained over time" ], "echoes": [ { "date": "2023-08-23", "setting": "UCSF, a different participant (47-year-old woman, brainstem stroke 2005): 253-channel ECoG, 78 words per minute, 1,024-word vocabulary, voice and avatar", "model_version": "253-channel ECoG; bidirectional RNN decoders", "outcome": "recurred", "source": "Metzger SL, et al. (2023). A high-performance neuroprosthesis for speech decoding and avatar control. Nature. https://pmc.ncbi.nlm.nih.gov/articles/PMC10826467/" }, { "date": "2024-05-20", "setting": "UCSF, same participant (the full text at PMC11554235 describes a 36-year-old Spanish-English bilingual with anarthria and severe spastic quadriparesis after a bilateral pontine stroke, a 128-electrode ECoG array and the BRAVO trial, and refers to initial work with this participant): bilingual (Spanish-English) sentence decoding without specifying the target language", "model_version": "ECoG with deep-learning and bilingual natural-language models", "outcome": "recurred", "source": "Silva AB, et al. (2024). A bilingual speech neuroprosthesis driven by cortical articulatory representations shared between languages. Nature Biomedical Engineering. doi:10.1038/s41551-024-01207-5" } ] } } }, "provenance": { "kind": "legacy_record", "basis": "Preserved from the existing AIBD record; external sources were not rechecked.", "record_fields": [ "account", "replication" ], "source_refs": [] }, "review_state": "record_mapped", "coverage": "partial" } }, "summary": { "total": 91, "addressed": 91, "known": 35, "reviewed": 0, "record_mapped": 35, "not_assessed": 56, "complete": false, "by_state": { "value_present": 35, "not_reported": 0, "unresolved": 0, "not_assessed": 56, "not_applicable": 0, "withheld": 0, "source_unavailable": 0 } } } }