Tag Archives: machine learning

A DNA Tool to Predict the Future

A common fear among parents is the risk of their children being diagnosed with serious illnesses such as heart disease and cancer. Fortunately, a DNA tool created by Michigan State University has been shown to accurately predict people’s height and bone density. This tool could also potentially assess people’s risk for illnesses such as heart disease and cancer.

Figure1. A new tool, created by Michigan State University, capable of predicting height. Used with permission, © 2018 Science Daily.

The Potential:

This amazing tool uses an algorithm to predict human traits including height, bone density and even the level of education an individual might achieve, entirely based on one’s genome. Lead investigator of the study Stephen Hsu mentions, “This is only the beginning.” There are plans in motion to apply this tool to predict more relevant matters like heart disease, diabetes and breast cancer. These are major diseases that impact a large amount of people. Heart disease, for instance, is the second leading cause of death in Canada. About 1 in 12 Canadian adults live with diagnosed heart disease alone! Hence, this tool could greatly impact and advance healthcare. Doctors would be able to step in during the early stages of illnesses to prevent or delay them!

Creating and Testing:

Stephen Hsu and his team’s research used machine learning, where a computer learns from data, to analyze the entire genome of around 500 thousand adults in the United Kingdom. The genome’s obtained from the United Kingdom Biobank, an international resource for health information, were used to put the algorithm to work. The algorithm evaluated each individual’s DNA and taught the computer to identify unique differences. “The algorithm looks at the genetic makeup and height of each person, learns from each person, and ultimately produces a predictor that can determine how tall they are from their genome alone,” said Hsu.

The computer was able to accurately predict everyone’s height within about an inch. However, predictions for bone density and educational achievement were not as accurate. Nevertheless, they were accurate enough to identify outliers in the population who were at risk of having problems in school and low bone density, which is associated with osteoporosis.

Future Plans:

Hsu’s team plans to continue to improve the algorithm by obtaining larger, more diverse data sets for the computer to learn. This would increase the algorithm’s capability to understand and interpret the combination of genes responsible for risks against specific diseases.

With increasing technology, the bright future of healthcare is considerably closer than originally believed. “What was once thought to be 5 to 10 years out, is now a lot closer when it comes to this type of work,” Hsu mentioned.

Figure 2. Collecting DNA using mouth swabs. Used with permission, © 2015 InfoLaw.

It is quite possible in the near future that a simple swab of the cheek could save families millions of dollars in treatment costs and more importantly, the lives of their children through early interventions of genetically based diseases.

Trevor Shen

Will Artificial Intelligence One Day Rule the World?

So what is artificial intelligence (AI) anyways? Just as humans scientifically exhibit what we call “natural learning”, artificial intelligence gives machines the same essence of knowledge enhancement called “machine learning”. AI allows machines to think about previous experiences and predict future unseen data without humans explicitly programming them to do so. So does that mean one day they will accumulate enough learning material to outsmart the creators? If so, scientists hypothetically refer to this as “super intelligence”.

Image result for what is ai

Photo Credit: ProSysCom

https://www.prosyscom.tech/innovation-future/what-is-ai-and-machine-learning-in-construction-our-definitive-answer-innovation/

Google AI: Alphago phenomenon

Go is an abstract strategy board game, created by humans more than 2,500 years ago. AlphaGo AI developed by Google, astonishingly defeated the world’s best Go players. AlphaGo not only defeated them, but impressively “trained on 160,000 games recorded from top Go players and then on 30 million more games it played against itself” said Brenden Lake. Clearly,  the AI was able to train repeatedly at speeds unachievable by humans, learned from them and invented moves that no human has played before.

Photo Credit: Fiuba Consulting Club

http://fiubaconsulting.club/go-game-board/

So is AlphaGo considered a type of “super intelligence”?

Not quite. AlphaGo’s knowledge was encapsulated to the game Go, and only Go. It has no other signs of expanding into solving chess, checkers or anything else. Therefore we can state that although AlphaGo have officially outclassed humans at Go, it is not a “super intelligence”.

From within our phones, laptops to Google Homes and Amazon Alexa, AI is in our daily lives. We as humans tend to be skeptical of the unknown, and it is no different with AI. The lack of understanding of the general public is sole reason these topics to arise. Although AI does allows a machine to learn tasks without a human’s guidance, but at the end of the day it comes down to highly statistical algorithms. In the nutshell of AI, it’s primary function is to recognize relationships amongst enormous data to predict relevant, measurable answers. Hence over time, as new data comes in and joins the statistical model, its able to “learn” and perform even more accurate predictions for the future.

“Machine learning and artificial intelligence are tools. They can be used in a right or a wrong way, like everything else.” – Brenden Lake

Ultimately,  AI’s capability to learn at speeds beyond any human doesn’t implicitly mean it will secretly learn other intelligence as well to outsmart us. There simply is no definition of “natural learning” for humans yet, therefore “machine learning” also cannot possibly surpass such threshold and go beyond our capabilities. Hence, AI will only learn and be exceptional in the field they are designed to thrive in. Even if AI one day is able to learn outside of it’s design specifications, it doesn’t necessarily mean it will be successful either.  As we are only understand the surface of AI’s true powers, its difficult to predict what other possibilities exist, but taking over the world is improbable.

– Jason Gan