Method, system and computer-readable storage medium for mycelium biocomposite property prediction and method, system and computer-readable storage medium for processing myceliumelectrography signals

Abstract

The present disclosure is directed to a method, system and computer-readable storage medium for mycelium biocomposite property prediction. The method can include selecting target properties for a MyBC in one or several of its development phases. The method can further include inputting the target properties into a trained ML model, wherein the output of the trained ML model comprises a prediction of development phases key features such as formulation, organic, material and digital that has the highest probability to lead to the target properties. Furthermore, the present disclosure is directed to a method, system and computer-readable storage medium for processing myceliumelectrography signals.

Description

[0001] Method, system and computer-readable storage medium for mycelium biocomposite property prediction and method, system and computer-readable storage medium for processing myceliumelectrography signals

[0002] Introduction

[0003] In recent years the field of Machine learning (ML) has seen applications in the field of chemical engineering, pharmacology, organic chemistry, battery development and others related such as, but not limited to, selected patents e.g. WO2017094899A1 and WO2022203734A1 .

[0004] Mycelium-based biocomposites (MyBCs) represent an emerging class of sustainable materials that harness the natural growth of fungal mycelium - the root-like network of fungal threads (hyphae) - to create versatile and environmentally friendly materials. These composites are formed when mycelium grows through and binds together organic substrates such as agricultural waste products, creating a natural biological matrix.

[0005] The process begins by selecting suitable fungal species and combining them with organic materials that serve as both nutrients and structural components. During the growth phase, the fungal mycelium network expands through the substrate, breaking down organic matter and simultaneously creating an interconnected network that acts as a natural binding agent. Once the desired growth is achieved, the material is processed (typically through heat treatment) to halt fungal activity, resulting in a stable composite material.

[0006] What makes MyBCs particularly interesting is their unique combination of properties: they are biodegradable, renewable, and can be grown into virtually any shape. Their properties can be tuned by varying parameters such as the fungal species used, substrate composition, growth conditions, and post-processing methods. Applications range from packaging materials and building insulation to acoustic panels and even furniture, offering a promising alternative to traditional petroleum-based materials.

[0007] However, one of the key challenges in MyBC development has been the lack of systematic understanding of the relationship between growth parameters and final material properties, making it difficult to consistently produce materials with specific desired characteristics. This has led to increased interest in applying advanced monitoring techniques and machine learning approaches to better understand and control MyBC development. Machine learning could be implemented to better understand MyBC development, but the lack of set methodology that can measure the internal and external development of the fungi the information used in the ML models and applications would not yield expected results. For their part, mycelium-based biocomposites have been developed, marketed, and sold in selected regions of the world. Their production is made from curated and well selected material. Nevertheless, the lack of automation and broader knowledge of the fungal development, its industrial production is hindered, impeding the reduction of the production cost.

[0008] The lack of a systematic acquisition of technical, physical, chemical and / or biological internal and external data of the behavior and or development of fungi in several environments and / or under other perturbations (atmospheric, electric, magnetic, etc.) is a big challenge for the correct deployment of ML models to accelerate the understanding and / or control of development and / or processing of fungi. This is a strong feature needed for an industrially scaled production and is hindering the global production of a sustainable material.

[0009] Filamentous fungi exist as a multicellular, multinuclear network of hyphae, and communication-mediated cell fusion is an important aspect of colony development at each stage of the life cycle. The communication of hyphae is fundamentally based on an alternating sequence of electrical and chemical events at the microscopic level. In the species, encoding and transmitting information through electrochemical and ionic signals is facilitated by the active involvement of proteins and unique fungi constituents. In a hyphal colony (mycelium), cells engage in cell-to-cell communication, synaptic-like transmissions by modulating chemical transmitters or ionic currents which traverse their membranes, thus generating an electromagnetic field. Likewise, they undergo chemotropic growth toward each other until they make physical contact. Once contact is established, both cells will initiate the process of cell fusion, in which cell walls are remodeled, plasma membranes fuse, and ultimately the two cells become one with a shared cytoplasm. This process results in the hallmark interconnected colony associated with filamentous fungi. Cell fusion between hyphae within an individual colony reinforces network architecture and influences the flow of resources throughout the colony. Hyphal fusion can also occur between fungal colonies and can result in one of two outcomes. First, two genetically similar colonies can fuse together, resulting in shared resources and collaboration instead of competition. Second, if two colonies that are genetically dissimilar at non-self-recognition loci undergo fusion, the fused cells are compartmentalized and program.

[0010] The challenge lies in the possibility of exploiting these induced fields. It is plausible that if a substantial number of hyphae cells reach a critical mass and activate synchronously, the resultant electric and magnetic field should be detectable. To simplify this complex situation, the concept of the equivalent current dipole (ECD) is introduced. The ECD is a singular dipole that replicates the electric field produced by all individual dipoles in a specific mycelium region, summarizing the net effect of all microscopic currents. This approximation is widely adopted in the field of neuroelectromagnetism.

[0011] When stimulation extends beyond a focal mycelium region, each region is simulated by an equivalent current dipole, leading to a source distribution. The primary task is correlating active and pure mycelium regions, where other non-fungal materials could disrupt the generated electric fields. This process is intricately linked to the establishment of physical structures, such as boundaries that enclose regions with specific physical properties, like conductivity, meaning pure mycelium as close to the substrate as possible, but only mycelium within the contacts. The resulting model, termed the volume conductor model, forms the physical foundation for source analysis, subdivided into two major challenges. The first involves calculating the electric potential generated by known electrochemical sources within the MyBC at specific surface points, known as the forward myceliumelectrographic (fMEG) problem. For more than six decades, the forward Electroencephalographic (fEEG) problem has been a focus of research since Wilson and Bayley attempted to quantify the interaction between brain activity and the potentials they generate at the species' surface. We make use of this theory and principles to achieve our objectives.

[0012] The reconstruction of the sources responsible for the recorded values, termed the corresponding inverse EEG problem, demands a high level of detail, achievable only through numerical models. This is a milestone for the MEG techniques, but we have a lot of help from the developed fields. These models fall into two categories: boundary element models (BEMs) and finite element models (FEMs). While BEMs efficiently represent major tissue compartments, they lack the capacity to detail anatomical information within compartments, such as hyphae folding. On the contrary, FEMs capture these intricate details but are labor-intensive and computationally demanding. To advance understanding, a thorough mathematical analysis is vital for recognizing underlying phenomena and identifying the limitations of developed algorithms. This analysis tests the impact of variable modifications on the system's output and provides deeper insight into physical behavior. It also validates numerical models.

[0013] Consider the scenario where experimental field studies on a species are conducted, but for enhanced understanding, a mathematical model interpreting the measurements is desired. This understanding is pivotal for the lacking knowledge about the behavior of fungal systems within the organic soil ecosystems, which absorb more than one third of the CO2 in the planet.

[0014] The presented methodology provides a solution to the systematic acquisition of technical, physical, chemical and / or biological internal and external data of the behavior of fungi in several environments and / or under other perturbations (atmospheric electric, magnetic, etc.) and in several phases of its development from conception, fungal organic development, material production and characterization. By introducing a methodology with steps to describe and / or label the composites in contact with the fungi, the specimen itself, and / or the container, and / or the environment and / or the physical and chemical tests where its being developed and researched. This by making use of manually and / or automatically introduced data from a user and / or computer system, and / or usage of analog and / or digital signals production and / or detection, as well as actuators. The methodology obtains relevant data of the internal development of the fungi by means of electromagnetic and / or conductive signal.

[0015] This methodology also introduces a sub-methodology which is crucial for the process which is the internal sensing of the fungi development within the environmental matrix in-situ. By implementing and collecting conductive experiments and detecting and producing electromagnetic fields, this in the right conditions and steps, the data collected allows for the methodology and the researcher to recognize important features of the mycelium-based biocomposite. Likewise external by means of sensors and / or spectrographs, such as, but not limited to, cameras or gas sensors provide external data collection.

[0016] The material development is done by killing the fungi and creating a set of posterior processes for the further development and / or characterization of the now considered material. The embedding and / or implementation of the data by the described methodological production, analysis and / or correlation will allow for the application of ML approaches for the prediction, automation and / or optimization of features which will be targeted values. By training the models with the generated data prediction of features could be traced back to composition formulations, environmental conditions and / or posterior processing steps. These activities will be crucial and paramount for the industrialization of the mycelium-based biocomposite production at large scale and with an adaptational advantage to environmental changes, availability of materials and / or other production bottlenecks.

[0017] The mycelium biocomposite (MyBC) is defined as a multi-composite body for the effects of the presented disclosure. The composite is a mixture of a plurality of environmental matrices and the fungi in specific ratios of the total mass in the specific geometric volume containing the mass within one or more atmospheric spaces. By methodically collecting data from but not limited to the first four phases, a digital twin of the same phases can be formed and used in ML models and applications. That is, a digital collection of labels, values, notes, experimental data, images, in the form of bits and that can be read and modified by computer systems. MyBC is then a biocomposite which can be methodologically produced and that can have a collection of digital values describing the conception, design, production, and performance throughout the defined lifespan.

[0018] We define Fungi as any organism of the kingdom Fungi. The Kingdom Fungi is limited to eukaryotes that form chitinous, resistant propagules (fungal spores) and chitinous cell walls and that lack undulipodia (that is, are amastigote or immotile) at all stages of their life cycle. Of the 1.5 million species of fungi estimated to exist, about 60,000 have been described; most are terrestrial, although a few truly marine species are known. The four fungus classifications are Chytridiomycota (chytrids), Zygomycota (bread molds), Ascomycota (yeasts and sac fungi), and Basidiomycota (basidiomycetes) (club fungi). The sexual or asexual evolution and reproduction (or spore production) of the organism takes place in an environmental matrix.

[0019] An environmental matrix (EM) is an organic or inorganic material or a mixture of such where the fungi develop and colonizes. The plurality of the environmental matrix of different natures and in different amounts is defined as environmental matrices (EMs). EMs can be tested and / or processed before their contact with fungi in a plurality of ways. The parameters that define an EM are the material nature (organic, inorganic, compound, alloy, etc.), physical form, size of the particles composing the material, a homogeneity factor, water content, amount of mass and / or ratio in comparison to the total mass and others. We define Geometrical Space (GS) to the confined space in which the fungi and or the EMs are contained. This space can be replicated and the material can be of any sort in which the MyBC shows development and can be of inorganic and / or organic origin. The volume generated by the confinement will be filled up to some extent with fungi and / or EMs. The coordinates (x, y, z), distances (cm, ft) and / or materials as well as locations are some, but not limited, of the information to be acquired from GS.

[0020] We define Atmospheric Space (AS) to the space where one and / or a plurality of geometrical space are confined. A plurality of GS can be within an AS. As well, one AS can contain a plurality of ASs. The coordinates (x, y, z), distances (cm, ft) and / or materials as well as locations are some, but not limited, of the information to be acquired from AS.

[0021] We define Sensing (Ss) as the set of values obtained in-situ and / or ex-situ where set of fungi and EMs are interacting. These measurements can include periodic and / or transient capacitance, sheet resistance, resistivity, conductivity, impedance and / or other more complex methods. Ss can further include a periodic spectral image capture with and / or without light exposure in a wide spectrum from the infrared to the ultra-visible using cameras and spectrographs. The stored values can form a time series dataset which is attributed to the MyBC and later analyzed to obtain characteristic physical values, such as relative humidity, electrograms, spectrograms of specific discretized information through data correlation. Ss as described can be performed by the sensors mentioned below.

[0022] We define as myceliumelectrography (MEG) in the present invention as a diagnostic technique that measures and records the electrical activity generated by fungal mycelium networks. It operates on the principle that fungal hyphae, like neural cells, produce electrical signals during their growth and communication processes. The technique adapts principles from electroencephalography (EEG) to fungal systems, providing insights into mycelium behavior and development through the detection and analysis of characteristic wave patterns. In some embodiments of the present invention, a processed MEG report is overlaid over a raw MEG report to permit a machine learning embedded model or technician to clearly apply and / or analyze the activity reported and provide a property prediction. The present invention provides the ability to select short overlapping epochs where the results of artifact removal from each epoch is stitched together with the result from the next and previous epoch. This allows for sensing and obtaining important information of the MyBC in-situ in the organic development phase.

[0023] We define the atmospheric conditions (AC) as the set of physical values characterizing the plurality of AS where the fungi and the EMs are. These can include, but not limited to, Temperature, Atmospheric Pressure, CO2 concentrations, O2 concentrations, CO concentration, NH2 concentrations, and other small aromatic molecules concentrations and / or Airborne particles. These measurements can be periodic and / or transient and might form a time series data which is then linked to the specific set AS set.

[0024] We define Post-Processing (PP) to the physical and / or chemical processes that the set of fungi and EMs go through once the decision to terminate the organic development phase of the MyBC is decided and / or achieved by a set numerical conditionals or thresholds. These can include, but not limited to, placing in an oven for a period of time and with specified temperature at a time to a achieve the MyBc in the material phase (i.e. when the fungi are dead), compressed using a mechanical and / or hydraulic press in the organic and / or material phase, coated with organic and / or inorganic materials in the form of solutions during the organic and / or material development phases, and / or mixing with other organic and / or inorganic materials to achieve a post-formation texture in the organic or material phase, as well as reshaping by means of placing into new geometrical spaces alone and / or with other materials.

[0025] We define Characterization (Ch) to the set of physical and / or chemical tests taking place before and / or after the post-processing processes to the MyBC. These can include, but not limited to, testing using a universal testing machine where the values for friction, scratch, linear wear, rotating wear, tensile & compression, fatigue, adhesion, and / or sense of touch are tested. As well as, but not limited to, flammability tests, impact tests, chemical and / or biological tests. The obtained values may be stored, and later used to obtain specific values attributed to the MyBC and used in specific analysis and / or in models for property prediction.

[0026] We define as Embedding (Em) to the process where the final set of digital values attributed the specific set of fungi, EMs and / or MyBC which can include, but limited to, labels, experimental data, fitted values, categorical and / or a plurality of attributed specific values are stored into a MyBC dataset. The Em can include the sampling, transformation, editing and / or consolidation of the plurality of MyBC datasets to be later transformed accordingly to the chosen ML model and / or application. These transformations can include normalization, labeling, dimensionality reduction, averaging, standard deviations, and / or other statistical and analytical transformations such as mathematical / physical model simulations. The physical body of the MyBC might then be removed from the plurality of AS, labeled, and passed to another phase which can include but are not limited to selling, further experimentation, degradation studies and or stored.

[0027] We define as Machine Learning (ML) a field of computer science that enables computer systems to learn and improve from experience without being explicitly programmed. It uses statistical techniques and algorithms to analyze patterns in data, allowing systems to identify relationships, make predictions, and perform tasks by learning from examples rather than following pre-set rules. In embodiments of the present invention, ML is the phase where the data consolidated in Em and / or external data from other sources is combined to form one or a plurality of ML embedded models with selected features and values to predict. A plurality of training and testing methodologies can be applied to achieve a higher score in the ML prediction model values which will translate in more accurate predictions of properties and / or formulation descriptions.

[0028] The present methodology presents the transformation of the fungi and EMs known as elements into a MyBC. This by undergoing selected development phases. These development phases can comprise formulation (containment in GS & AS), organic (Ss which contains MEG, AC and / or other biological values), material (containing PP, Ch) and / or Digital (Em and / or ML). The produced MyBC in physical form has then a set of stored digital values describing the process for its future analysis, replication, and industrialization.

[0029] To advance understanding, a thorough mathematical analysis is vital for recognizing underlying phenomena and identifying the limitations of developed algorithms. This analysis tests the impact of variable modifications on the system's output and provides deeper insight into physical behavior. It also validates numerical models. Consider the scenario where experimental field studies on a species are conducted, but for enhanced understanding, a mathematical model interpreting the measurements is desired. This understanding is pivotal for the lacking knowledge about the behavior of fungal systems within the organic soil ecosystems, which absorb more than one third of the CO2 in the planet. This same can be applied to industrial set-ups of mass production, like seen in the brewing, or wine industries.

[0030] The present disclosure develops novel tools and protocols for reducing the number of experiments in MyBC research by a factor of 3, and, more generally, for boosting the pace of material discovery and industrialization of marketable applications by a factor of > 50. Machine Learning (ML) stands out as a promising approach that could lead to a paradigm shift in the way we do MyBC R&D and production, enabling us to overcome the major challenges dealing with a vast number of variables and large quantity of data for property prediction and / or optimization of features.

[0031] MyBC production and / or research is a complex multivariable problem, where very different properties, such as performance, life-cycle analyses, safety, cost, environmental effects, and energetic and / or resource issues, are contained. Furthermore, the overall MyBC circular economy should eventually be included from the crop planning and recollection, production, and logistical stage via the long usage phase to the final reuse and recycling processes. The disclosed workflow, however, relies heavily on a forward trial-and-error approach and is largely, but not limited to, materials, but it could be used for production with a few modifications from the included methodology: selecting materials, manufacturing equipment, standardized reactors, and finally assessing performance. Even considering only these aspects, there are >1O100possibilities to formulate MyBC materials and postprocess them almost an infinite number of possibilities for choosing the set of fungi and EMs manufacturing parameters and dozens of possible formats and climates, which is far greater than what a human brain can handle. This makes difficult the emergence of inverse design tools enabling the prediction of the MyBC properties needed for a given performance target and format.

[0032] ML assists to efficiently solve the parameters and data challenges of MyBC formulations as well as assist the R&D of MyBC technologies beyond MyBC production, such as but not limited to, mushroom fruiting, in-situ research, and / or industrial upscaling of production and / or its parametrization, ending by calculating the sustainability factors of such processes. This methodology looks to introduce measurement and design standards in MyBC R&D, and production combined with systematic data collection, analysis and disclosure, the identification of the most suited descriptor(s) for a certain ML model, or the determination of the associated error, among others. The vast majority of mycelium composite products are single fungal species with specific EMs formulations. The field of machine-learning (ML) has advanced rapidly in being able to predict the physical and perceptual properties of materials, but applications in the field of mycology and specific for mycelium-based composite formulations are largely ignored.

[0033] MyBC models in the art focus on perceptual similarity of EMs for predictions while ignoring other factors. For example, certain existing approaches focus on storing and providing human acquired data on properties of EMs such as prices, quantities, quality, water content, drying pressure, etc.

[0034] SUMMARY

[0035] Aspects and advantages of embodiments of the present disclosure will be set forth in part in the following description, or can be learned from the description, or can be learned through practice of the embodiments.

[0036] The present disclosure is directed to a computer- implemented method for MyBC property prediction. The method can include selecting target properties for a MyBC in one or several of its development phases. The method can further include inputting the target properties into a trained ML model, wherein the output of the trained ML model comprises a prediction of development phases key features such as formulation, organic, material and digital that has the highest probability to lead to the target properties.

[0037] Alternatively, the method can include selecting a feature or composition material in the development phases wherein the selected development phases can be input into a trained ML and wherein the output of the trained ML model comprises a prediction of MyBC properties that have the highest probability to be the result of the selected development phases.

[0038] A method for training the trained ML model can include generating a plurality of datasets by following the methodology in the development phases of the MyBC. The datasets can be generated in part by humans inputting manual data, digital and / or analog sensors (such as, but not limited to, atmospheric sensors, or testing machines), digital and / or analog actuators (such as, but not limited to, hydraulic press, ovens), computer generated calculations (such as, but not limited to, fittings, simulations) and data previously stored in servers (such as, but not limited to, clouds or physical memory devices) The fungi, EMs and / or MyBC properties can include, but are not limited to, physical, chemical, nutritional, mechanical, energetic, pharmaceutical, conductive, financial, thermal, catalytic, structural, and soil remediating properties. For example, if MyBC is used as a building material, the most important properties are their physical properties which can comprise for example density, stiffness, and thermal conductivity. Predicting these properties or the parameters for the development phases that lead to certain desired properties can be very valuable for industrial applicability of MyBC.

[0039] The method can include manufacturing MyBC materials with targeted properties by using development key features. For example, targeted properties can be input into the trained machine learning model and the predicted development phase key features that are identified by the trained machine learning model to have a highest probability to lead to the target properties can be used for manufacturing a MyBC material. Alternatively, selected development phase key features input into the trained machine learning model can be used to manufacture a MyBC material with properties that have a highest probability to be the result of the selected development phase key features according to the prediction of the trained machine learning model.

[0040] The trained ML model can include a machine-learned embedding model to generate a respective embedding for each MyBC. The method can include processing, by the computing system, the embeddings and the MyBC data set with a prediction model to generate one or more property predictions for the fungi and / or EMs of the plurality of fungi and / or EMs. In some implementations, the one or more property predictions can be based at least in part on the embeddings and the MyBC data. The method can include storing, by the computing system, the one or more property predictions, past formulations of the MyBC, biological, chemical, composition geographic, economic, demographic and / or financial data.

[0041] In some implementations, the MyBC data can describe a respective concentration of each fungi and / or EM in the EMs. The MyBC data can describe a composition of the EMs. The ML prediction model can include a deep neural network. In some implementations, the ML embedding model can include a ML graph neural network. The prediction model can include a characteristic-specific model configured to generate predictions relative to a specific characteristic. The one or more property predictions can be based at least in part on one or many synergistic coefficients of one or more fungi and or EM of the plurality of fungi and or EMs. The synergistic coefficient can be a measure of competition, when its value is negative, and of enhanced development when positive. In some implementations, the one or more property predictions can include one or more physical property predictions. The one or more property predictions can include a tensile strength prediction. The one or more property predictions can include a catalytic property prediction. In some implementations, the one or more property predictions can include a proteinic fruiting yield property prediction. The one or more property predictions can include a post-processing between target property prediction.

[0042] In some implementations, the one or more property predictions can include an industrial property prediction. The one or more property predictions can include a thermal property prediction. The prediction model can include a weighting model configured to weight and pool the embeddings based on the MyBc data, and the EMs data can include concentration, biological, and financial data related to the plurality of fungi and / or EM of the EMs.

[0043] In some implementations, the method can include obtaining, by the computing system, a request from a requesting computing device for a MyBC with a requested property, determining, by the computing system, the one or more property predictions satisfy the requested property, and providing, by the computing system, the MyBC data to the requesting computing device. The one or more property predictions can be based at least in part on a fungi and or EMs synergistic interaction property. In some implementations, one or more property predictions can be based at least in part on receptor threshold activation data or value.

[0044] Another example aspect of the present disclosure is directed to a system, in particular a computing system. The computing system can include one or more processors and one or more non-transitory computer readable media that collectively store instructions that, when executed by the one or more processors, cause the computing system to perform operations. The operations can include obtaining respective fungi and or EMs data for a plurality of fungi and or EMs, MyBC data and / or development phases data. In some implementations, the MyBC data can include concentrations for each respective fungi and or EM of the plurality of the fungi and or EMs. The operations can include respectively processing the respective fungi and or EM data with an embedding model for each of the plurality of fungi and or EMs to generate respective embeddings for each MyBC. The operations can include processing the embeddings and the MyBC data with a machine- learned prediction model to generate one or more property predictions. The one or more property predictions can be based at least in part on the embeddings and the MyBC data. The operations can include storing the one or more property predictions. Another example aspect of the present disclosure is directed to one or more non- transitory computer readable media that collectively store instructions that, when executed by one or more processors, cause a computing system to perform operations. The operations can include the mentioned above methodology and their key values and features.

[0045] Furthermore, the present invention introduces an innovative MEG system and method. In some embodiments, a processed MEG report is overlaid over a raw MEG report, enabling a clearer visualization of mycelial activity in a specie. This invention offers the capability to select short, overlapping epochs where the results of artifact removal from each epoch are seamlessly integrated with the results from adjacent epochs. This integration, or "stitching," is performed using a weighted average method, with the weight being proportional to the distance to the center of each epoch. For example, if an epoch length of two seconds is selected with a one-second increment, artifact removal is performed on successive two- second intervals, producing overlapping clean results. The overlapping results are then combined through a weighted average for each channel, producing a continuous and artifact-free final result.

[0046] In some embodiments, multiple sensors are distributed and attached to the geometric volume where one species or several species are contained, connected to a device via surface conductivity means. This device amplifies the signals and records the mycelium’s electrical activity. The electrical activity results from the summation of neural-like activity across numerous hypha cells, which generate small electric voltage fields. The aggregate of these electric voltage fields creates a reading which the sensors detect and record. An MEG is a superposition of multiple simpler signals. Typically, the amplitude of an MEG signal ranges from 100 nanovolts to 500 millivolts, measured with more precise instruments.

[0047] The MEG is performed to monitor mycelium during critical processes and other physical perturbations. In MEG, one or multiple sensors (standard positions for at least 1) are distributed inside and at the surface. These sensors are referenced, for example by their position in relation to an even geometrical distribution of the area where the species’ mycelium is contained. Contrary to EEG, where a labelling of the distribution is already by convention, in here the labeling is adapted according to the geometry.

[0048] The MEG records mycelium waves from one or more amplifiers using various combinations of sensors called montages, created to provide a clear picture of the spatial distribution of the MEG across the geometric volume. In bipolar montages, consecutive pairs of sensors are linked, and adjacent channels have one sensor in common. In a referential montage, various sensors are connected to one input of each amplifier, and a reference sensor is connected to another input of each amplifier. This montage collects signals at an active sensor site and compares them to a common reference sensor, good for determining the true amplitude and morphology of a waveform.

[0049] Locating the origin of electrical activity (“localization”) is critical in analyzing the MEG. Localization in bipolar montages is achieved by identifying “phase reversal,” where two channels within a chain point in opposite directions. In a referential montage, channels may show deflections in the same direction. The electrode with the largest upward deflection represents the maximum negative activity in a referential montage.

[0050] Some patterns in MEG indicate a tendency toward mycelial perturbations. These include spikes, sharp waves, and spike-and-wave discharges. Spikes and sharp waves in specific areas, such as the surface or bottom, indicate that maturing processes might be taking place. Primary generalized mycelial phases are suggested by spike-and-wave discharges spread over the geometry, especially if they start simultaneously.

[0051] Brief Definitions Used in MEG Context:

[0052] "Amplitude" for example, refers to the vertical distance measured from the trough to the peak of the neuron-like mycelium population and its activation synchrony.

[0053] "Analogue to digital conversion" refers to converting analogue signals (like MEG) into a digital form for computer processing.

[0054] "Artifacts" for example are electrical signals detected on the species’ surface originating from non-mycelial sources. These can include specie-related artifacts (Magnetic and electrical responses, atmospheric responses, biological responses) and technical artifacts (electrical interference, sensor movement).

[0055] “Electrode” in the MEG context refers for example to a conductor used to establish electrical contact with a circuit, here referring to mycelium surface sensors.

[0056] “Morphology” refers to the shape of the waveform, determining the nature of mycelium wave patterns.

[0057] “Montage” means the placement of the sensors for MEG monitoring.

[0058] “Neural network algorithms” refer to algorithms identifying sharp transients. “Noise” refers to any unwanted signal modifying the desired MEG signal.

[0059] “Processed reports” refers to statistical and or mathematically modified raw data with information that can be overlaid in the raw data.

[0060] Despite advancements, current software systems often fail to accurately distinguish true fungal mycelial signals from artifacts, necessitating expert assessment. Therefore, a need exists for an enhanced method and system that allows for a clear comparison of raw and processed MEG data, aiding in the precise analysis of a species' mycelial activity.

[0061] In some embodiments, the present invention provides a MEG system and method that overlays a processed MEG report over a raw MEG report to enable a researcher or technician to clearly see the activity reported in a species. This invention offers the capability to select short overlapping epochs, wherein the results of artifact removal from each epoch are seamlessly integrated with the outcome of the adjacent epochs. This integration can be executed in various ways, but a preferred method involves combining the signals from the two epochs using a weighted average, where the weight is proportional to the ratio of the distance to the epoch centers.

[0062] For instance, an epoch length of two seconds might be chosen, with an increment (epoch step) of one second. Artifact removal, utilizing Blind Souce Separation (BSS, see below) and other techniques, is conducted on a set of channels for the first and second second, yielding a two-second length "clean result." Subsequently, artifact removal is applied to the second and third seconds, producing an overlapping clean result. The overlap occurs in the second second of the record. For each channel, the weighted average of the two overlapping results is calculated to produce a final outcome devoid of discontinuities. In the segment of the second closer to the center of the first epoch, the value from the first epoch is given greater emphasis, and similarly, the portion closer to the center of the second epoch is weighted more heavily.

[0063] Experts in the relevant field will acknowledge that various epoch lengths or steps may be chosen while progressing through the record. Additionally, alternative stitching techniques could be employed to achieve the desired outcome. This advanced approach in MEG technology enhances the clarity and precision of mycelium activity analysis, catering to the intricate requirements of contemporary biological research.

[0064] One aspect of the present invention is a method for filtering artifacts from a MEG signal. The method involves generating a MEG signal from a system equipped with a multitude of electrodes, an amplifier, and a controller adapted to perform the described method processor specifically designed for mycelial studies. This method also entails the transformation of the MEG signal from a set of channels into multiple epochs. Each epoch within this array can have a duration of no more than two seconds at a predetermined time and / or an increment not exceeding one second half of that predetermined time.

[0065] The core of this method is the filtration of artifacts from each of these epochs. This is achieved using a blind source separation algorithm. In particular, Blind Source Separation (BSS) is a signal processing technique that separates a set of mixed signals into their original source components without prior knowledge of the mixing process or the sources themselves. The term "blind" refers to the fact that both the source signals and the way they were mixed are unknown. The algorithm works by identifying statistically independent components within the mixed signals, making it particularly useful for removing noise and artifacts from complex signal recordings. In contexts like myceliumelectrography (MEG), BSS can separate genuine mycelial electrical activity from various forms of interference and noise. The output of this process is a series of clean epochs, each representing a clearer and more accurate snapshot of mycelial activity.

[0066] The method can incorporate a step where these clean epochs are meticulously combined to form a cohesive and processed MEG recording. This composite recording offers a more precise and artifact-free representation of mycelial activity, thereby enhancing the accuracy and reliability of the data for further analysis and interpretation. This innovative approach is instrumental in advancing the study and understanding of mycelial behavior and communication.

[0067] Yet another aspect of the present invention is a method for filtering artifacts from a MEG signal using a blind source separation algorithm or z-normalized Euclidean distance algorithms. The method entails generating a MEG signal from a system comprising a multitude of electrodes, an amplifier, and / or a processor specifically tailored for mycelium studies. This method also involves transforming the MEG signal from a set of channels into a series of epochs. Each epoch in this series can have a duration that does not exceed two seconds and / or a predetermined amount of time.

[0068] The crucial part of this method is the filtration of artifacts from each epoch using for example a blind source separation algorithm, effectively isolating and removing any unwanted disturbances from the mycelial signal. This process results in the creation of a series of clean epochs, each providing a more accurate representation of mycelial activity.

[0069] Furthermore, these clean epochs are then meticulously combined to produce a coherent and processed MEG recording. This refined MEG recording offers a high-fidelity representation of mycelial activity, crucial for accurate analysis and research.

[0070] An additional aspect of the present invention is a system for filtering artifacts from a MEG signal. This system can comprise one or more electrodes, an amplifier, a controller, and / or a display unit. The electrodes are responsible for generating MEG signals, which are then amplified by the amplifier connected to each electrode. The controller, linked to the amplifier, is adapted to execute the inventive method, in particular generating a MEG recording from these signals. The display, connected to the processor, is used for showcasing the MEG recording.

[0071] In this system, the controller is adeptly configured to transform each MEG signal from a set of channels into multiple epochs. It then removes artifacts from each epoch using the blind source separation algorithm, resulting in a series of clean epochs. These clean epochs are subsequently combined to create a processed MEG recording, which is then displayed, offering a high-resolution and artifact-free view of mycelial activity.

[0072] The method and system together represent a significant advancement in the field of mycelial research, providing researchers with enhanced tools for accurate data collection and analysis, particularly in the context of understanding complex mycelial networks and behaviors.

[0073] The method includes generating a MEG signal from a system that can comprise multiple electrodes, one or more amplifiers, one or more processors and / or a controller designed for mycelium studies. The method can also involve transforming the MEG signal from a set of channels into a series of epochs. Furthermore, the method can include filtering artifacts from each epoch using an artifact removal algorithm to generate a series of clean epochs. These clean epochs can then be combined to create a processed MEG recording.

[0074] Another aspect of the present invention is a method for filtering artifacts from a MEG signal by selecting specific epoch times and increments. This method can involve generating a MEG signal for a species from a system equipped with multiple electrodes attached to the mycelium, an amplifier, and / or a controller. The method includes choosing an appropriate epoch time length and increment. Artifacts can then be filtered from each epoch using an artifact removal algorithm to create clean epochs. A weighted average can be assigned to each of these clean epochs, ensuring a smooth transition and overlap, ultimately leading to a processed MEG recording without discontinuities.

[0075] Additionally, the present invention encompasses a system for filtering artifacts from a MEG signal. This system can include electrodes, one or more processors, a controller, and / or a display. The electrodes are responsible for generating MEG signals from mycelium. The processor, connected to these electrodes, is tasked with generating a MEG recording from the signals. The display, linked to the processor, showcases the MEG recording. The processor in this system is specially configured to select epoch time lengths and increments, filter artifacts from each epoch using an artifact removal algorithm to generate clean epochs, assign weighted averages to these epochs, and combine them in such a way that they overlap seamlessly, thus producing a continuous and processed MEG recording without any discontinuities.

[0076] This method and system significantly advance the field of mycelium research, offering refined tools for accurate data analysis and enhancing the understanding of complex mycelial behaviors and interactions.

[0077] Other aspects of the present disclosure are directed to various systems, apparatuses, non- transitory computer-readable media, user interfaces, analogous and or digital sensors, and electronic devices. These and other features, aspects, and advantages of various embodiments of the present disclosure will become better understood with reference to the following description. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate example embodiments of the present disclosure and, together with the description, serve to explain the related principles.

[0078] DETAILED DESCRIPTION OF FIGURES

[0079] Figure 1 A depicts an example evolutionary approach 100, which can be used for generating a database of new MyBC with predicted properties. The proposed MyBC can have MyBC data 102 for each respective proposed MyBC. The MyBC data 102 can be processed by the machine-learned property prediction system 104 to generate predicted properties 106 for the proposed MyBC. The predicted properties 106 can then be processed by an objective function 108 to decide whether an addition to the corpus of top performers 110 should be made or whether to discard. A random and or selective mutation can be made, and the process can begin again. The evolutionary approach 100 can aid in generating a large database of useful mixtures of MyBc to be available for screening by a human practitioner or further ML models for use in a variety of products and industries. The MyBC data 102 can comprise of relational or non-relational databases which correlate data acquired during the development phase one or a plurality of MyBC and such data can be accompanied with external data pertinent or correlated with the MyBC data subject to analysis and or prediction and the MyBC data 102 can be collected by computer-controller sensors and or actuators as well as manually by a researcher. Corpus of top performers means the output collections with higher predicting scores in the prediction of the determined property, i.e. top 5 formulations which achieve flexibilities of lower than certain flexibility coefficient. It can also be predefined values from mathematical equations such as, but not limited to, density (kg / m3) as well as subjective synergistic coefficients generated by the researcher. The objective function 108 means / can comprise of various loss functions such as, but not limited to mean squared error, likelihood loss, cross entropy loss, hinge loss, and / or various other loss functions which score according to the chosen loss functions. Gradient descent techniques can be used to iteratively update the parameters over a number of training iterations. In some implementations, performing backwards propagation of errors can include performing truncated backpropagation through time and perform a number of generalization techniques (e.g., weight decays, dropouts, etc.) to improve the generalization capability of the models being trained.

[0080] Figure 1 B depicts an example reinforcement learning approach 150, which can be used for model optimization. Similar to the evolutionary approach 100, the reinforcement learning approach 150 can begin with MyBC data 102 of a proposed MyBC being processed by a machine-learned property prediction system to generate predicted properties 106. The predicted properties 106 can then be processed by an objective function 112 to provide an output to a machine-learning controller 114 to provide a proposal to the system. In some implementations, the machine-learning controller can include a recurrent neural network. In some implementations, the reinforcement learning approach 150 can aid in refining the parameters of the machine-learned models disclosed herein.

[0081] Figure 2 depicts a block diagram of an example property prediction system 200 according to example embodiments of the present disclosure. In some implementations, the property prediction system 200 is trained to receive a set of input data 202, 204, 206, and 208 descriptive of MyBCs and, as a result of receipt of the input data 202,204, 206, and 208, provide output data 216 that includes one or more property predictions descriptive of predicted properties of a MyBC, such as but not limited to, formulations, temperature values, structural modeling, etc. The input data 202,204, 206, and 208 may also be incomplete. For example information about certain properties can be missing from single or multiple datasets. The input data 202,204, 206, and 208 may also be non-uniform. One of the advantages of ML is that incomplete data can still be used for training a respective ML model. Thus, in some implementations, the property prediction system 200 can include one or more embedding model(s) 212 that are operable to generate MyBC embeddings, and a machine-learned prediction model 214 that is operable to generate one or more property predictions 216. Property prediction systems 200 can include two-stage processing of input data to generate one or more property predictions 216. For example, in the depicted system 200, the input data can include MyBC data with respective MyBC data 202, 204, 206, and 208 for each MyBC, in which the MyBC data can be descriptive of an N number of MyBC data 210 descriptive of the composition of a MyBC of the N number of MyBC. The system 200 can process the MyBC data with one or more embedding model(s) 212 to generate one or more embeddings to be processed by the machine-learned prediction model 214. In some implementations, the embedding model 212 can include a graph neural network (GNN) to generate one or more graphs. In some implementations, the MyBC data can be processed such that the respective MyBC data related to each individual MyBC can be processed separately such that each embedding can represent a singular MyBC sample.

[0082] The embeddings and the mixture data 210 can be processed by the machine- learned prediction model 214 to generate one or more property predictions 216. The machine- learned prediction model 214 can include a deep neural network and / or various other architectures. Moreover, the property predictions 216 can include various predictions related to various properties associated with the MyBC. For example, the property predictions 216 may include physical property predictions, such as a tensile property prediction to later be used for creating a specific composite. Furthermore, in this implementation, the first MyBC 202, the second MyBC 204, the third MyBC 206, ..., and the nth MyBC 208 can be of the same or different concentrations in the theoreticized MyBC. The system may weigh the one or more embeddings based on concentration of the MyBC. The weighting can be completed by the embedding model 212, the machine-learned prediction model 214, and / or a third separate weighting model. Figure 3 depicts a block diagram of an example property prediction system 300 according to example embodiments of the present disclosure. The property prediction system 300 is similar to property prediction system 200 of Figure 2 except that property prediction system 300 further includes three initial predictions. More specifically, the depicted system 300 includes three initial predictions being made before the overall property predictions 330 are generated. For example, the system 300 can make individual MyBC predictions 310, MyBC composition property predictions 322, and MyBC interaction property predictions 324, which can all be factored into the overall property predictions 330. The system 300 can begin with obtaining input data 310, which can include MyBC data descriptive of a mixture with a set of MyBC. The input data can be processed by a first model to generate MyBC specific predictions 310, and in some implementations, the predictions 310 can be formulation specific predictions. The formulation predictions 310 may be weighted based on the concentration of the Fungi and or EMs in the specified geometric space and the predictions of the various MyBC may be pooled. The output of the first model can then be processed by a second model 320, which can include two sub-models. The first sub-model can process the data and output composition of the new formulation with same or new fungi and or EMs in same and or different concentrations for the geometric space and the specific property predictions 322 associated with the overall composition in percentage of total mass or by specific concentrations of the MyBC. The second sub-model can process the data and output interaction specific property predictions 324 associated with predicted interactions in the MyBC and / or predicted extrinsic interactions. The three initial predictions can be processed to generate an overall property prediction 330 based on each of the initial predictions to allow for a better understanding of the MyBC. For example, each individual MyBC may have their own respective flammability properties, while certain compositions may lead to some MyBC properties being more prevalent. Moreover, interaction properties of various individual and sets of MyBC may alter, enhance, or dilute certain flammability properties. Therefore, each initial prediction can provide insight to how the overall MyBC may break, combust, etc.

[0083] Figure 4 depicts a block diagram of an example computing system 400 that performs property predictions according to example embodiments of the present disclosure. The system 400 can include a user computing device 402, a server computing system 430, a telemetry system for sensors and actuators 470, and / or a training computing system 450 that are communicatively coupled over a network 490. The user computing device 402 can be any type of computing device, such as, for example, a personal computing device (e.g., laptop or desktop), a mobile computing device (e.g., smartphone or tablet), a gaming console or controller, a wearable computing device, an embedded computing device, or any other type of computing device. The user computing device 402 includes one or more processors which could be of central processing unit (CPU), quantum processing unit (QPU), tensor processing unit (TPU), or a combination thereof. 412 and a memory 414. The one or more processors 412 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, a FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. The memory 414 can include one or more non-transitory computer-readable storage mediums, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. The memory 414 can store data 416 and instructions 418 which are executed by the processor 412 to cause the user computing device 402 to perform operations.

[0084] The sensor modulated device 470 includes one or more processors 472 which could be of central processing unit (CPU), quantum processing unit (QPU), tensor processing unit (TPU), or a combination thereof and a memory 474. The one or more processors 472 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, a FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. The memory 474 can include one or more non- transitory computer-readable storage mediums, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. The memory 474 can store data 476 and instructions 478 which are executed by the processor 472 to cause the user computing device 472 to perform operations. The embedded or external sensors 480, included those specified for the attached invention MEG 484 and or actuators 482 are controlled by the combination of processor 472, memory 474 to perform actions and using the communication module 484 which send data 486 over the network 490 and exchange instructions 488 for the processor 472. The communication 484 can be through several of the possible communication protocols possible (e.g., TCP / IP, HTTP, SMTP, FTP).

[0085] The server computing system 430 includes one or more processors 432 and a memory 434. The one or more processors 432 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, a FPGA, a controller, a microcontroller, CPU, QPU, TPU or a combination.) and can be one processor or a plurality of processors that are operatively connected. The memory 434 can include one or more non-transitory computer-readable storage mediums, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. The memory 434 can store data 436 and instructions 438 which are executed by the processor 432 to cause the server computing system 430 to perform operations. The user computing device 402 and / or the server computing system 430 can train the models 420 and / or 440 via interaction with the training computing system 450 that is communicatively coupled over the network 490. The training computing system 450 can be separate from the server computing system 430 or can be a portion of the server computing system 430.

[0086] The training computing system 450 includes one or more processors 452 and a memory 454. The one or more processors 452 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, a FPGA, a controller, a microcontroller, CPU, QPU, TPU or a combination) and can be one processor or a plurality of processors that are operatively connected. The memory 454 can include one or more non-transitory computer- readable storage mediums, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. The memory 454 can store data 456 and instructions 458 which are executed by the processor 452 to cause the training computing system 450 to perform operations. In some implementations, the training computing system 450 includes or is otherwise implemented by one or more server computing devices.

[0087] The training computing system 450 can include a model trainer 460 that trains the machine- learned models 420 and / or 440 stored at the user computing device 402 and / or the server computing system 430 using various training or learning techniques, such as, for example, backwards propagation of errors. For example, a loss function can be backpropagated through the model(s) to update one or more parameters of the model(s) (e.g., based on a gradient of the loss function). Various loss functions can be used such as mean squared error, likelihood loss, cross entropy loss, hinge loss, and / or various other loss functions. Gradient descent techniques can be used to iteratively update the parameters over a number of training iterations.

[0088] The network 490 can be any type of communications network, such as a local area network (e.g., intranet), wide area network (e.g., Internet), or some combination thereof and can include any number of wired or wireless links. In general, communication over the network 490 can be carried via any type of wired and / or wireless connection, using a wide variety of communication protocols (e.g., TCP / IP, HTTP, SMTP, FTP, MQTT), encodings or formats (e.g., HTML, XML, ASCII), and / or protection schemes (e.g., VPN, secure HTTP, SSL, MQTT, NodeRED).

[0089] Figure 5 depicts a flow chart for generating MyBC training data 500 for the previously described ML prediction model. As previously specified certain elements of the methodology are crucial elements of a MyBC and others are digital values of the production process. A selected and labeled set of fungi 501 which could be of any of the four classifications in any stage of their development (spore, hyphae, and / or others), a set of a plurality of labelled and characterized environmental matrices (substrates, additives, organic and inorganic materials) 502, a set of geometric spaces (blocks, cylinders, abstract) 503, a set of atmospheric conditions (dry, humid, dynamic, 32 °C, etc.) 504 and / or a set of postprocessing steps (drying, compressing, bending, covering, etc.) 505 produce a unique MyBC 506. This MyBC 506 will be transformed into a unique set of objective functions and features 507 linked to this unique MyBC 506. The digital set of data of 506 and objective functions and features 507 are stored in a user computer device or memory storage 508. The data can be transformed by mathematical routines to achieve multidimensional representations as well as being subjected to dimensionality reductions. The analysis of principal component analysis (PCA), independent component analysis (ICA), matrix profiling (MP) as well as frequency and wavelet analysis can be performed for the time correlated data. PCA can for example be understood as a dimensionality reduction technique that transforms correlated variables into a set of linearly uncorrelated variables called principal components. In particular it identifies the directions (principal components) in which the data varies the most, allowing to reduce the number of dimensions while retaining the most important patterns in the data. ICA can for example be understood as a method that separates a multivariate signal into additive, statistically independent components. Unlike PCA, which finds uncorrelated components, ICA finds components that are as statistically independent as possible. MP can for example be understood as a technique for analyzing time series data that computes a distance profile between a subsequence and every other subsequence in the time series. The resulting matrix profile provides a measure of similarity between all possible subsequences in the data. MP can for example make use of z- normalized Euclidian distance as a key measure, wherein z-normalization can comprise subtracting the mean from a subsequence and dividing it by its standard deviation. Final embedded databases can later be of numerical, vectorial, text or binary nature or a combination thereof.

[0090] Figure 6 depicts the flowchart of the production methodology for producing MyBC and digital data for training embedded ML models 600. The digital data can be seen as digital twin to the physical MyBC. The development phases of the MyBC are the unique formulation, such as but not limited to, fungi, Environmental Matrices, Geometrical Spaces, Atmospheric Conditions and / or Post Processes, the organic phase 610 in which data is collected through the described methodology Sensing 621 which could also made use of myceliumelectrography 622 for in situ MyBC data collection. After the organic phase 610 reached conclusion the last development phase takes place, material phase 630 which is a physical, chemical and or biological process that the material undergoes before, during and after characterization 631 methodologies and processes such as but not limited to universal testing, flammability testing, etc. The digital phase of the MyBC contains the Embedding phase 640 which is the collection and storing of analog and digital data as well as categorical and other types of data and further mathematical analysis or fitting to determine objective functions and or features from the development phases 610, 620 and or 630. Lastly, machine learning is the phase where a single and / or a plurality of sets of MyBC data from the embedding phase 640 can be combined with external data or other MyBC data to train embedded ML models for predicting the properties of MyBC as shown in figures 1A and 1 B and be following the methodologies from figures 1 , 2, 3 and 4.

[0091] Figure 7 shows a raw or original MEG report 700 that could be the result of a myceliumelectrography 622. The original MEG report 700 has a plurality of channels, shown at the Y axis 705 of the report. The X-axis of the report is time. The original MEG report 700 has not been subjected to artifact reduction. The original MEG report contains artifacts from various sources such as, but not limited to, electrode cables, electromagnetic interference, and the like. However, the MEG may also have certain activity that a researcher is looking for from the MEG report in order to accurately analyze the MyBC activity. For example, the activity shown 710 at a time 90,000 s may represent a certain stage of aerobic activity for the MyBC that is important to the researcher of an embedded machine learning model.

[0092] Figure 8 is an illustration of a processed MEG report 800 of the original MEG report 700 of Figure 7 that has undergone artifact reduction and the stitching of epochs in order to recreate the MEG report, however, the epochs do not overlap, resulting in lost information and / or discontinuities. The processed MEG report 800 has a plurality of channels, shown at the Y axis 805 of the report. The X-axis of the report is time. As shown in 810, the processed MEG report 800 at time 90,000 is quite different in appearance than the original MEG report 700 at time 90,000. This is primarily due to the stitching of epochs to recreate the MEG report; however, if a researcher was only looking at the processed MEG report 800, the researcher would not be aware of the true activity at time 90,000.

[0093] Figure 9 is an illustration of an MEG report 900, based on the MEG report 700 of Figure 7, in which channels have been removed for a clearer illustration of channels. The illustration of the combined MEG report 900 only has three channels in order to clearly illustrate the invention; however, those skilled in the pertinent will recognize that the combined MEG report 900 could have six, twenty, twenty-seven and any number of channels without departing from the scope and spirit of the present invention.

[0094] A flow chart for a method 1000 for displaying MEG data is shown in Figure 10. At block 1001 , an original MEG report is generated from an MEG signal. The original MEG report is generated from an MEG machine comprising a plurality of electrodes and processors. The original MEG report comprises a first plurality of channels. At block 1002, the original MEG signal is partitioned from a set of channels into epochs of which each has a predetermined duration length and an over-lap increment. At block 1003, artifact reduction is performed on the epochs to generate artifact-reduced epochs. At block 1004, the artifact-reduced epochs are combined with overlapping adjacent epochs for a continuous MEG recording to generate a processed continuous MEG report. The stitched, overlapping epochs and continuously processed MEG report is displayed on a display screen, preferably a monitor. The stitched, overlapping epochs and continuous processed MEG report is not missing timeframes from stitching or creating discontinuities in the MEG report, which is read by a researcher or technician. All of the MyBC activity remains, since the epochs overlap. The MyBC activity is preferably spikes, sharp waves, spike and wave discharges, artifacts, and the like.

[0095] Figure 11 is a flow chart of a preferred method 1100 for displaying MEG data. At block 1101 , an original MEG report is generated from an MEG signal for a MyBC from a machine preferably comprising electrodes attached to the sample, an amplifier and a processor. At block 1102, the original MEG signal is partitioned from a set of channels into a plurality of epochs. Each of the plurality of epochs has an epoch duration length and an overlap increment. At block 1103, a first artifact reduction is performed on the plurality of epochs to remove electrode artifacts. At block 1104, a second artifact reduction is performed on the plurality of epochs to remove EMs artifacts. At block 1105, a third artifact reduction is performed on the plurality of epochs to remove external signal artifacts. At block 1106, the plurality of epochs are combined to overlap, wherein each epoch of the plurality of epochs overlaps an adjacent epoch to form a processed continuous MEG report. At block 1107, a processed continuous MEG recording is generated from the combined epochs.

[0096] The artifact removal algorithm is preferably a blind source separation algorithm. The blind source separation algorithm is preferably a CCA (canonical correlation analysis), MP (matrix profile) and ICA (Independent Component Analysis) to transform the signals from a set of channels into a set of component waves or "sources." The sources that are judged as containing artifacts are removed and the rest of the sources are reassembled into the channel set. The clean epochs are preferably combined using a weighted average and the weight of the weighted average is preferably proportional to the ratio of the distance to an epoch center.

[0097] Figure 12 is an isolated view of adjacent unprocessed epochs 1201 and 1202. Epoch 1201 has an overlapping portion 1203 and epoch 1202 has an overlapping portion 1204. In this example, the overlapping portions 1203 and 1204 are approximately 150 seconds in length. Thus, overlapping portions 1203 and 1204 represent the same timeframe for raw MEG recording.

[0098] Figure 13 is an illustration of adjacent processed epochs 1301 and 1302. Artifact reduction has been performed on these epochs 1301 and 1302. Processed epochs 1301 and 1302 represent the same timeframe as unprocessed epochs 1201 and 1202 in figure 12. Thus, epoch 1301 is the result of artifact reduction of unprocessed epoch 1301 , and epoch 1302 is the result of artifact reduction of unprocessed epoch 1202 in figure 12. Processed epoch 1301 has an overlapping portion 1303 and processed epoch 1302 has an overlapping portion 1304. Thus, overlapping portions 1303 and 1304 represent the same timeframe for the processed MEG recording. Further, overlapping portion 1303 is the same timeframe as overlapping portion 1203 and overlapping portion 1304 is the same timeframe as overlapping portion 1204. Further overlapping portions 1203, 1204, 1303 and 1304 represent all the same timeframe.

Claims

C l a i m s1 . A method, in particular a computer-implemented method, for mycelium biocomposite (MyBC) property prediction and / or manufacturing, the method comprising:- selecting, in particular by one or more processors, target properties for a MyBC;- inputting, in particular by the one or more processors, the target properties into a trained machine learning model;- generating, by the trained machine learning model, a prediction of development phase key features that have a highest probability to lead to the target properties.

2. A method, in particular a computer-implemented method, for mycelium biocomposite (MyBC) property prediction and / or manufacturing, the method comprising:- selecting, in particular by one or more processors, development phase key features of a MyBC;- inputting, in particular by the one or more processors, the development phase key features into a trained machine learning model;- generating, by the trained machine learning model, a prediction of properties for the MyBC that have a highest probability to be the result of the selected development phase key features.

3. The method, according to one of the preceding claims, further comprising manufacturing a MyBc with the target properties using the development phase key features.

4. The method, according to one of the preceding claims, wherein the development phase key features comprise: selection and concentration of fungi, environmental matrices composition, geometrical spaces, atmospheric conditions and / or post-processing parameters.

5. The method, according to one of the preceding claims, further comprising generating the trained machine learning model by training a machine learning model using a plurality of datasets of MyBC samples, the datasets comprising in particular data input manually by humans, data from sensors, data from actuators and / or computer-generated calculations.

6. The method, according to the preceding claim, further comprising generating the datasets of MyBC samples using cameras, gas sensors and / or myceliumelectrography (MEG) sensors configured to detect internal development of fungi within an environmental matrix in-situ through electromagnetic and / or conductive signal detection.

7. The method, according to claim 5 or 6, wherein generating the trained machine learning model comprises processing the datasets by the one or more processors with an embedding model to generate respective embeddings, in particular for each MyBC sample.

8. The method, according to one of the preceding claims, wherein the target properties comprise physical properties, chemical properties, mechanical properties, thermal properties, structural properties and / or biological properties.

9. A system for mycelium biocomposite (MyBC) property prediction, comprising one or more processors and memory-storing instructions that, when executed by the one or more processors, cause the system to perform operations comprising:- selecting, in particular by one or more processors, target properties for a MyBC;- inputting, in particular by the one or more processors, the target properties into a trained machine learning model;- generating, by the trained machine learning model, a prediction of development phase key features that have a highest probability to lead to the target properties; or- selecting, in particular by one or more processors, development phase key features of a MyBC;- inputting, in particular by the one or more processors, the development phase key features into a trained machine learning model;- generating, by the trained machine learning model, a prediction of properties for the MyBC that have a highest probability to be the result of the selected development phase key features.

10. A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:- selecting, in particular by one or more processors, target properties for a MyBC;- inputting, in particular by the one or more processors, the target properties into a trained machine learning model;- generating, by the trained machine learning model, a prediction of development phase key features that have a highest probability to lead to the target properties; or- selecting, in particular by one or more processors, development phase key features of a MyBC;- inputting, in particular by the one or more processors, the development phase key features into a trained machine learning model;- generating, by the trained machine learning model, a prediction of properties for the MyBC that have a highest probability to be the result of the selected development phase key features.

11. A method, in particular a computer-implemented method, for processing myceliumelectrography (MEG) signals, the method comprising:- generating, by a plurality of electrodes of one or more MEG sensors, MEG signals from mycelium within a geometric volume;- partitioning, by one or more processors, the MEG signals from a set of channels into a plurality of epochs, wherein each epoch has a predetermined duration length and an overlap increment with adjacent epochs;- performing, by the one or more processors, artifact reduction on each epoch using a blind source separation algorithm to generate artifact-reduced epochs;- combining, by the one or more processors, the artifact-reduced epochs using weighted averaging to form a processed continuous MEG recording.

12. The method, according to the preceding claim, wherein the weighted averaging is proportional to a ratio of distance to epoch centers of adjacent epochs.

13. The method, according to claim 11 or 12, wherein performing artifact reduction comprises performing a first artifact reduction to remove electrode artifacts, performing a second artifact reduction to remove environmental matrices artifacts and performing a third artifact reduction to remove external signal artifacts.

14. The method, according to any one of claims 11 to 13, wherein the predetermined duration length is no more than two seconds and / or the overlap increment is no more than half of the predetermined duration length.

15. The method, according to any one of claims 11 to 14, wherein the blind source separation algorithm comprises transforming the MEG signals into a set of component waves using canonical correlation analysis (CCA), matrix profile (MP) and independent component analysis (ICA), identifying and removing component waves containing artifacts and reassembling remaining component waves into the set of channels.

16. The method, according to any one of claims 11 to 15, wherein combining the artifact-reduced epochs comprises identifying overlapping portions between adjacent epochs, applying weighted averaging to the overlapping portions and stitching the epochs together to create a continuous processed MEG recording without discontinuities.

17. A system for processing myceliumelectrography (MEG) signals, comprising:- a plurality of electrodes of one or more MEG sensors configured to generate MEG signals from mycelium within a geometric volume;- one or more amplifiers connected to the plurality of electrodes;- one or more processors connected to the one or more amplifiers; wherein the one or more processors are configured to:- partition the MEG signals from a set of channels into a plurality of epochs, wherein each epoch has a predetermined duration length and an overlap increment with adjacent epochs;- perform artifact reduction on the epochs using a blind source separation algorithm to generate artifact-reduced epochs;- combine the artifact-reduced epochs using weighted averaging to form a processed continuous MEG recording.

18. The system, according to the preceding claim, wherein the plurality of electrodes are arranged in a bipolar montage configuration wherein consecutive pairs of electrodes are linked and adjacent channels share one electrode or a referential montage configuration wherein multiple electrodes are connected to one amplifier input and a reference electrode is connected to another amplifier input.

19. The system, according to claim 17 or 18, wherein the one or more processors are further configured to detect mycelium wave patterns including spikes, sharp waves and / or spike-and-wave discharges.

20. The system, according to any one of claims 17 to 19, wherein the one or more processors are further configured to identify phase reversals in bipolar montages to locate origins of electrical activity.

21. The system, according to any one of claims 17 to 20, wherein the one or more amplifiers are configured to detect MEG signals ranging from 100 nanovolts to 500 millivolts and amplify the MEG signals for processing by the one or more processors.

22. The system, according to any one of claims 17 to 21 , wherein the plurality of electrodes is distributed in a geometrical arrangement within and on a surface of the geometric volume and the electrodes are referenced by their position in relation to the geometrical arrangement.

23. A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:- generating, by a plurality of electrodes of one or more MEG sensors, MEG signals from mycelium within a geometric volume;- partitioning, by one or more processors, the MEG signals from a set of channels into a plurality of epochs, wherein each epoch has a predetermined duration length and an overlap increment with adjacent epochs;- performing, by the one or more processors, artifact reduction on each epoch using a blind source separation algorithm to generate artifact-reduced epochs;- combining, by the one or more processors, the artifact-reduced epochs using weighted averaging to form a processed continuous MEG recording.

Images