Real-time neural spike detection

A method is described for real-time detecting and classifying of a characteristic signal, such as a neural spike, and forwarding information for further processing if it meets certain criteria. A system (e.g., an on-chip system implanted in a subject's cranium with limited processing power) receives an electrical biological signal. The system filters the signal to generate a filtered signal and fits the filtered signal to a model. The system identifies a set of fit values based on the model, the set of fit values comprising a plurality of sample amplitude values and a respective plurality of time values. Based on the fit values, the system computes a set of characteristic values. The system compares the characteristic values to a corresponding set of threshold values. Based on the comparison, the system determines whether the received biological signal corresponds to a neural spike and, if a spike is detected, forwards on information.

The abstract as published by the patent office, from US2021012909A1. It describes the invention; it is not a statement of what is protected.

• Interface and electrodes

  Electrodes and arrays

• Signal processing

  Signal processing

• Decoding

  Neural decoding

What the filing appears to be about, read by published rules from its title, abstract and classification codes (A61B5/0031, A61B5/24, A61B5/291, A61B5/369, A61B5/37). The claims define what a patent protects, and they are not read here.

Interface and electrodes

Everything downstream works with what the contacts pick up. Their number, size and distance from the neurons set the ceiling on the signal. The problem behind it: more contacts, closer to neurons, that still record after years in tissue that reacts to them.

Signal processing

A decoder is only as good as the features it is given, and recordings drift and pick up noise from movement, muscle and stimulation. The problem behind it: stable features from signals that change from day to day, computed fast enough to act on.

Decoding

This is the step that turns brain activity into a command or a word. Its accuracy and speed are what a user experiences. The problem behind it: accuracy that holds over months without frequent recalibration, for more than one user and one task.

Why the layer matters to the field, not a judgement of this filing. A patent family shows that protection was sought. It does not show that the technology was built, is used in a product, or works in people. The whole stack

One invention, published 14 times in 7 offices. Each publication opens in Espacenet.

• United StatesUS2021012909A114 Jan 2021US12688941B221 Jul 2026 · grant
• CanadaCA3146297A121 Jan 2021CA3146297C15 Aug 2023 · grant
• International (PCT)a route to several countries, not a countryWO2021011296A121 Jan 2021
• AustraliaAU2020315738A110 Feb 2022
• South KoreaKR20220025119A3 Mar 2022KR102469238B118 Nov 2022 · grant
• European Patent Officea route to several countries, not a countryEP3996593A118 May 2022EP3996593A426 Jul 2023EP3996593B17 Jan 2026 · grantEP3996593C07 Jan 2026
• JapanJP2022532958A20 Jul 2022JP7158622B221 Oct 2022 · grant

1. 12 Jul 2019First filingEarliest priority, at United States. The family is counted in this year.
2. 9 Jul 2020ApplicationsFiled at United States, Canada, International (PCT), Australia, South Korea, European Patent Office, Japan.
3. 22 Jul 2020Owner recordedUnited States, on US2021012909A1: assignment; owner named: NEURALINK CORP., CALIFORNIA.
4. 14 Jan 2021First publicationUS2021012909A1, United States.
5. 19 Apr 2022Office recordSouth Korea, on KR20220025119A: notification of reason for refusal.
6. 7 Jun 2022Office recordJapan, on JP2022532958A: notification of reasons for refusal.
7. 21 Oct 2022Grant publishedJP7158622B2, Japan.
8. 18 Nov 2022Grant publishedKR102469238B1, South Korea.
9. 15 Aug 2023Grant publishedCA3146297C, Canada.
10. 7 Jan 2026Grant publishedEP3996593B1, European Patent Office.
11. 21 Jul 2026Grant publishedUS12688941B2, United States.
12. 6 Oct 2026Office recordEuropean Patent Office, on EP3996593A1: lapsed in a contracting state [announced via postgrant information from national office to epo] (ES, GR, HR, IS, NO, PL, RS, SM).

Dates as the offices published them. Office records are in the offices' own words and are a selection: refusals, lapses and withdrawals, and records that name an owner. Each concerns one member of the family, not the family as a whole, and another member in the same office may stand. Whether a patent is in force today is not assessed.

Organisation

Neuralink · 31 families identified

• Bibliographic data, family members and office records: European Patent Office, Open Patent Services (the DOCDB and INPADOC databases), read 7 Oct 2026.
• The record itself, with its claims and description: US2021012909A1 in Espacenet
• Counted as a BCI patent family: classified A61B5/369 (electroencephalography); filed by Neuralink Corp., an organisation already in the graph; describes neural hardware or methods without naming the field; cites a family already accepted. How families are found and judged