How Decentralized Artificial Intelligence and Machine Learning Change Medicine, Explained

These are the so-called 21st-century technologies.

Artificial intelligence is the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making and translation between languages.

Machine learning makes use of algorithms that learn how to perform tasks such as prediction or classification without explicitly being programmed to do so. In essence, the algorithms learn from data rather than being pre-specified.

There are various levels of machine learning and pure AI, beginning with:

• Supervised learning
• Unsupervised learning
• Deep learning

Each level takes vast amounts of data and can create relatable and usable information out of associations with a speed and accuracy comparable to, if not better than, humans. This is the power of artificial intelligence and machine learning.

Decentralized technology has several benefits.

Namely, data privacy and ability to create a collaborative atmosphere. This is also applicable to decentralized artificial intelligence. Machine learning models keep data safe and ensure privacy. How? They communicate back and forth and keep data on the other end user’s device. Additionally, once the models have continued learning and matured, they are open and accessible to everyone in the network. This way there’s no need in a centralized proprietary organization. It is important because currently, this authority is the ultimate one to determine the destiny of future discoveries.

Blockchain technology enables parties to interact.

These interactions are based on a set of agreed upon business rules. The rules can define payment transfers or generic rulesets called smart contracts. A decentralized network of peers host these rule sets to verify proposed transactions on a smart contract. The contracts can be used to define the manner in which various parties interact in a trustless, permissionless manner. And this type of network can lay the foundation for a platform. Data aggregation and deep learning models can be developed that would otherwise be too costly for centralized institutions to undertake.

In an age of mobile phones and tablets, these devices are the main computing devices for many people. Due to the fact that in today’s day and age, consumers are attached to their mobile device at the hip, there is a combination of rich user interactions and powerful sensors, leading to an unprecedented amount of data, which tends to be private in nature. Due to the sensitive nature of the data means there are risks and responsibilities to storing it in a centralized location. Therefore, the models taught on users data hold the promise of greatly improving usability by powering more intelligent applications.

Real Life Example

Image: Doc Ai

To see the explaining video go here.

There are multiple use cases.

In the world of medicine, there are projects, like Neuron, that have developed several interesting products in beta. These products will guide and teach users how to train their decentralized AI; in other words, how to train the trainer. Users will be able to see how to construct datasets of their health, and where and how to access these datasets.

Onboarding Module with Computer Vision

One product uses AI and ML to automatically fill in your physical statistics on the app just by taking a selfie. The Selfie2BMI module uses state-of-the-art Deep Neural Networks and optimization techniques to predict a variety of anatomic features including height, weight, BMI, age and gender from a face. Besides these vital anatomies, it also monitors 23 facial attributes like skin, receding hairlines, wrinkles, teeth and other attributes.

Blood Test Decoder

Another innovative use that Neuron has developed are deep conversational agents designed to enhance the post-blood test experience, enabling the user to discuss and answer any question on 400 blood biomarkers. It is trained on hundreds of thousands of medical documents and common FAQs to answer complex questions about blood results. The agent can personalize the conversation based on the user's age, gender and pre-conditions to provide relevant answers and educate using interactive content.

Genomics Test Decoder

A deep conversational agent designed to enhance the genetic counseling experience by providing answers from simple educational questions to complex personalized questions. It has a memory and remembers every visit and recommendation it has given. When it does not know, it goes to look for the answer in a massive dataset. It can also pass you on to its carbon-based colleague.

Medicine Decoder

An on-boarding module trained on medication dosage, side effects and other guidelines to answer personalized questions. It will be connected to the pharmacogenomic recommendation engine if genomic results are present.

It will make the medical experience more interactive and tailored to the user.

There are several key problems that decentralized AI tackles, and provide the user the opportunity to take back control of their healthcare.

The Integral Burden

These solutions can guide the participant in finding and collecting its own medical data. Most people do not have access to their medical information, they do not know where to start and if they do, their knowledge of computer science is limited. They also foster and support an open-source developer community to help innovate tools on their tool platform stacks, to facilitate the integration and collection of data and to present algorithms to interpret the data: a decentralized Kaggle for personalized biology.

Organic and Unbiased Data

The Blockchain is used as triple-entry accounting so we can track and authenticate the data sources. This allows to make great predictions and be able to audit the provenance of the data and perform data forensics and KYD or Know Your Data, processes.

Also for the biased data problem, the technology creates safer healthcare for the user. For example, data coming from randomized control trials are often riddled with bias. The highly selective nature of trials systematically disfavor women, the elderly, and those with additional medical conditions to the ones being studied; pregnant women are mostly entirely ignored.

Privacy Concerns

People may be hesitant to share their medical data over networks where strangers can be lurking. By decentralizing the data over all the users, it is cryptographically timestamped and becomes immutable. Furthermore, Neuron addresses HIPAA requirements by maintaining the information on the edge device, not in the cloud or on a centralized server.

Opportunities Ahead

While the healthcare industry remains far from creating a doctor in a machine, in medicine, there are generalists (GPs) and Specialists. Generalists are analogous to GAI (General AI), unreachable at this point of technological development. But specialists are like vertical AIs and closer to realization. The ABMS (American Board of Medical Specialties) lists more than 150 medical specialties and subspecialties. Neuron has the potential to be the leading player in this sphere.