AI Model For Neurodegenerative Diseases


I. ABSTRACT

Models of neural networks are receiving widespread attention as potential new architectures for computing systems. The models we consider here consist of highly interconnected networks of simple computing elements. A computation is performed collectively by the whole network with the activity distributed over all the computing elements. This collective operation results in a high degree of parallel computation and gives the network the potential to solve complex problems quickly.

II. INTRODUCTION
The prefectural concept of curing diseases has been a topic of discussion for generations. Every sort of cure has been implemented for various organs of the body for various diseases. From medicines to chemotherapy has shown success for the variety of diseases. But when it comes to neuro biological diseases, primarily the focusing on Neurodegenerative diseases,
success for searching a cure for these have had limited success through traditional means. Such diseases start a degeneration of certain areas of the brain. Since different parts of the brain have different functions, the degeneration causes a hindrance to perform said functions. The AI model proposed makes use of a neuromorphic chip, that can be trained to replace the damaged lobe of the brain and perform the function of the damaged part of the brain. The AI at this point of development has shown promise in possibly giving a solution for maybe the biggest disease of the human physiology.

Types of Neuro-Degenerative Diseases :

  1. Parkinson’s Disease
  2. Alzheimer’s Disease
  3. Huntington’s Disease
  4. Amyotrophic Lateral Sclerosis
  5. Spinal Muscular Atrophy
  6. Spinocerebellar Ataxia
  7. Prion Disease

III. HISTORY
Neurodegenerative diseases were first noticed around the end of the nineteenth century towards the twentieth century. Some of the more known diseases in this category include Parkinson’s disease, Alzheimer’s disease and to some lesser spread as the phantom’s limb.
Neuromorphic chips, an application of neuromorphic engineering, a concept first developed by Carver Mead in the late 1980s. In today’s world understanding how the morphology of individual neurons, circuits, applications, and overall architectures creates desirable computations, affects how information is represented, influences robustness to damage, incorporates learning and development, adapts to local change and facilitates evolutionary change.

AVAILABLE TREATMENT(Till date)

  • Increasing the Dopamine Signaling, using Antiviral Medication i.e., Amantadine or using Dopamine Agonists which stimulate receptors.
  • Inhibition of COMT(Catecholamine-o-methyltransferase) which degrades Dopamine production internally.
  • The most effective and cost-effective treatment includes; Deep Brain Stimulation using neuromorphic chip(an implantable device) that directly sends electrical signals to the Basal Ganglia which counteracts aberrant signaling.

IV. MODEL DESIGN

  • Hardware Model (Synapses)
    Neuromorphic chips are implemented using micro and nanoscale transistors. In neuromorphic chips transistors conventionally used as switches are also used as analog dials. To process some data out of it, synapses among the network need to be modified and its resemblance with human brain, enables it for same.
    The electronic neural network is based on a simplified model of a processing element(neuron) interconnected to many such neurons with resistors. Each neuron, i, gets current as input from another neuron, j through a resistor with conductance, Tij. The signal coming from neuron j to neuron i is given by
    Iij = Vj * Tij
    Where Iij is current flowing through node j to node i, Vj is the input voltage applied to node j and Tijis the conductance which represents the connection weight. All the currents from nodes 1, 2, 3, …n are summed at the input node(i) and output voltage of the neuron is the function of this total current. Typically, a neuron has a non-linear transfer function which can be a limiting threshold.
    Analog CMOS switches and x-y decoders, used in artificial designing of an array of synapses, are used to select a synapse cell for programming. Current summing effect has been observed at the input of a neuron. This array is designed using modular and registered structured elements and hence is easily expandable to larger networks. To Emulate neurons, a multichip system has been developed, called Neuro-Grid. Thus, at large, a chip can be trained artificially based on the variation of currents and thus modifying the synapses.
  • Software Model
    (Neuro-Computers and Application Software)
    Neural network is trained using Deep learning and the processors which can withstand parallel computing, called Neuro-Computers which are dedicated ones, for such purpose. Neuro-computer is feed with the formula required to train the network for a specific Neuro-Degenerative disease which by a processor, sends out signals to the analog circuit of transistors i.e., chip and thus the neural network gets trained.
    The Accelerator Boards play a very significant role in training with large data set required for the network to get trained about neuro-degenerative diseases. Also, it includes dedicated application software for better and direct mapping of respective patients with concerned healthcare department to cure using neuromorphic chips.

Architecture Model

V. Does it have an Application?

  • Hospitals and concerned healthcare divisions because they are far more efficient and a better way out to cure degenerative diseases whose prediction itself is a challenge to date.
  • To help the patients recuperate from the neurodegenerative diseases.
  • The AI model with neuromorphic chip finds application in space research as well. NASA is conducting research on Neuromorphic systems & Artificial Neural Membrane (ANM).

VI. SCOPE OF MODEL

  • The model creates a revolutionary change, preserving and saving one’s life, incorporating development and learning.
  • It is a subject that is interdisciplinary incorporating physics, computer science, electronic engineering, biology, and mathematics for designing neural artificial systems like eye-head systems, auditory processors whose design architecture and physical principles have basis on system of biological neural network.
  • With parallel computing, the network gets trained and so the respective chip developed in lesser comparable time, due to processors.

VII. CONCLUSION
The Stanford chip sounds like a significant approach to deal with Neuro-Degenerative diseases which can be trained with a large dataset and being analogous in nature, mimics the human brain more accurately than an implementation with digital technology. Concerned healthcare department mapped by a dedicated application software through which direct One-One connection would be developed between the patient of Neuro-Degenerative disease and the professional to inject the particularly developed chip for proper treatment.

VIII. REFERENCES

  • Jack L. Meador, Angus Wu, Novat Nintunze, and Pichet Chintrakulchai, “Programmable impulse neural circuits,”IEEE Trans. Neural Networks, Vol. 2, pp.101-109,1991.
  • Ramacher, U. et al. “SYNAPSE-1: a highspeed general-purpose parallel neurocomputer system “IPPS (774-781). 1995.
  • D.B. Schwartz, R.E. Howard, and W.E. Hubbard, “A programmable analog neural network chip,” IEEE J. Solid-State Circuits, Vol. 24, pp. 313–319, 1989.

.



My Personal Notes arrow_drop_up

Check out this Author's contributed articles.

If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to contribute@geeksforgeeks.org. See your article appearing on the GeeksforGeeks main page and help other Geeks.

Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.




Article Tags :

Be the First to upvote.


Please write to us at contribute@geeksforgeeks.org to report any issue with the above content.