METHODS FOR OPERATING A BIOLOGICAL NEURAL NETWORK COMPUTING SYSTEM WITH ADAPTIVE COMPUTATIONAL INTERFACES AND AUTOMATED CONTROL
A method for performing a task using a biological computing system is disclosed. The system comprises a biological neural network (BNN) including an in vitro neural cell culture, a stimulation unit configured to apply stimulation signals to the neural cell culture, a readout unit configured to measure readout signals from the neural cell culture, an automation controller, and pre-processing and post-processing units configured to respectively encode input signals into stimulation signals and decode output signals from measured readout signals. The system may further include a health control unit configured to regulate environmental conditions of the neural cell culture, including chemical, nutritional, and physical parameters. The automation controller may be configured to monitor and control system components and parameters through one or more communication interfaces, and to support user-provided instructions and/or automated control by computational agents, including natural language-based interfaces and machine communication protocols enabling bidirectional data exchange. In some embodiments, control instructions are generated and adapted dynamically based on system feedback, and automated control agents adjust encoding operations, decoding operations, and environmental and operational parameters of the biological computing system in response to monitored data.
1 . A method for using a biological computing system in furtherance of performing a task, the biological computing system comprising a biological neural network (BNN) comprising an in vitro neural cell culture, a stimulation unit adapted to apply a stimulation signal into the neural cell culture, an automation controller including a pre-processing unit adapted to encode an input signal into a stimulation signal for the stimulation unit to apply to the neural cell culture and a post-processing unit adapted to decode an output signal from a measured readout signal response of the neural cell culture, and a readout unit adapted to measure a readout signal produced by the neural cell culture in response to the stimulation signal, the method comprising:
receiving, by the pre-processing unit, an input signal to be processed by the biological computing system in furtherance of performing the task;
encoding, with the pre-processing unit, the input signal to generate a stimulation signal;
applying the stimulation signal to the neural cell culture with the stimulation unit
measuring, with the readout unit, a readout signal from the neural cell culture in response to the stimulation signal;
decoding, with the post-processing unit, an output signal from the measured readout signal from the neural cell culture;
using the decoded output signal from the post-processing unit in furtherance of performing the task.
2 . The method of claim 1 , wherein encoding, with the pre-processing unit, the input signal to generate a stimulation signal comprises applying filters, classifiers, or machine learning algorithms based on mathematical or statistical models (trained statistical models) to the input signal.
3 . The method of claim 2 , wherein the filters, classifiers, or machine learning algorithms based on mathematical or statistical models (trained statistical models) comprise one or more of a signal processing model, a computational model, signal filter, a signal classifier, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, a natural language model, a large language model, and a reservoir computing method.
4 . The method of claim 1 , wherein decoding, with the post-processing unit, an output signal from the measured readout signal comprises applying filters, classifiers, or machine learning algorithms based on mathematical or statistical models (trained statistical models) to the measured readout signal.
5 . The method of claim 4 , wherein the filters, classifiers, or machine learning algorithms based on mathematical or statistical models (trained statistical models) are selected among a signal filter, a signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, a natural language model, a large language model, and a reservoir computing method.
6 . The method of claim 1 , wherein the automation controller comprises a communication interface, and wherein a user provides control instructions through the communication interface for the automation controller to configure and/or to monitor one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit.
7 . The method of claim 6 , wherein the automation controller further comprises a natural language model or a large language model-based prompting interface that translates natural-language instructions into control instructions for the automation controller.
8 . The method of claim 6 , wherein the automation controller further comprises a natural language model or a large language model-based interface that converts data received from the communication interface into natural language or graphics.
9 . The method of claim 1 , wherein the automation controller further comprises a machine communication interface and wherein an AI agent writes and/or executes control instructions through the machine communication interface to configure and/or to monitor one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit.
10 . The method of claim 9 , wherein the machine communication interface comprises a Model Context Protocol interface or other structured communication protocol enabling bidirectional exchange of control instructions, stimulation signals and readout data with one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit.
11 . The method of claim 9 , wherein the AI agent autonomously generates control instructions comprising one or more of stimulation signals, configuration parameters and configuration commands, transmits them for execution to one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit via the machine communication interface, receives data information from one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit via the machine communication interface, and adapts subsequent control instructions based on the received data information.
12 . The method of claim 1 , wherein the biological computing system further comprises a health control unit adapted to adjust environmental parameters such as chemical additives, nutrients, temperature and/or CO 2 parameters of the neural cell culture.
13 . The method of claim 12 , wherein the automation controller further comprises a communication interface and wherein a user provides control instructions through the communication interface for the automation controller to configure and/or to monitor the health control unit and one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit.
14 . The method of claim 13 , wherein the automation controller further comprises a natural language model or a large language model-based prompting interface that translates natural-language user instructions into control instructions for the automation controller.
15 . The method of claim 13 , wherein the automation controller further comprises a natural language model or a large language model-based user interface that converts data received from the communication interface into natural language or graphics.
16 . The method of claim 12 , wherein the wherein the automation controller further comprises a machine communication interface and wherein an AI agent provides control instructions through the machine communication interface to configure and/or to monitor the health control unit.
17 . The method of claim 16 , wherein the machine communication interface comprises a Model Context Protocol interface or other structured communication protocol enabling bidirectional exchange of control instructions and/or monitoring data with the health control unit.
18 . The method of claim 16 , wherein the AI agent autonomously generates control instructions comprising one or more of stimulation parameters, stimulation commands, configuration parameters and configuration commands to adjust one or more of a chemical additive, a nutrient, a temperature and/or a CO 2 parameter of the neural cell culture, transmits them to the health control unit via the machine communication interface, receives data information from one or more of the health control unit, the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit via the machine communication interface, and adapts subsequent control instructions based on the received data information.
19 . The method of claim 18 , wherein the AI agent autonomously generates control instructions for the health control unit to command the release of a chemical additive or a nutrient to the BNN cell culture.
20 . The method of claim 12 , wherein the automation controller further comprises a communication interface, wherein a first AI agent provides first control instructions through a communication interface to configure and/or to monitor the health control unit to adjust one or more of a chemical additive, a nutrient, a temperature and/or a CO 2 parameter of the neural cell culture, and wherein a second AI agent provides second control instructions through a communication interface to configure and/or to monitor one or more of the pre-processing unit, the stimulation unit, the readout unit, and the post-processing unit.
21 . The method of claim 20 , wherein an AI Orchestrator agent provides control instructions through a machine communication interface to configure and/or to monitor the first and the second AI agents.
22 . The method of claim 20 , wherein the first AI agent autonomously generates control instructions for the health control unit to command the release of a chemical additive or a nutrient to the BNN cell culture.
23 . The method of claim 1 , wherein the task comprises learning to map a set of output signals to a set of input signals.
24 . The method of claim 1 , wherein the stimulation unit comprises multi-electrodes arrays, patch-clamps, light induced stimulation such as optogenetics systems, magnetic or electric fields, ion stimulation, focused laser light, optical tweezers, or mechanically induced stimuli through gravity or pressure changes.
25 . The method of claim 1 , wherein the readout unit comprises multi-electrodes arrays, patch-clamps, imaging systems, ion sensitive sensors, electrical or magnetic sensitive sensors, chemical sensors, or other sensors suitable to neuron cultures.