domingo, 16 de fevereiro de 2014

Watson and the Age of Cognitive Computing

In April 1964, IBM released the System/360, considered till today, after exactly 50 years ago, as a milestone in Computer Science. He transformed it in a unique way and paved the way for everything that we have seen in recent decades. At that time, IBM created a special division to market the new maiframe (read more about this special date on the blog of Daniel Raisch, totally dedicated to the 50th anniversary of the mainframe). Thirty years later, in the 90s, with increasing share of sales of the personal computer, the PC, IBM again created a division to market it.

They were two unique moments in the history of computing and of IBM. In 2007, IBM brought more great news, Watson. Designed by IBM laboratories, the idea is that the system is much more than a supercomputer. It represents a paradigm shift in the history of computing and makes way for a new computational model, the "cognitive computing".

Earlier this year, IBM created a new division to work with Watson, Watson Group, now with 2000 employees and announced investments of over 1 Billion dollars in the project. But ultimately, what IBM expects to Watson?

Cognitive computing

To better understand what this means, we must go back in the past. In the beginning of time, computing was characterized by "doing math quickly." The first computers were simply tabulating machines, ie, they did mathematical operations much faster than us. Until just over half of the twentieth century was how computers operated and that was what we expected of them.

During this period began to develop a little more intelligent machines which not only be able to repeat operations, but also to follow a set of instructions. It was the second time in modern computing, characterized by the ability to write computer programs that would later be executed by a computer.

The concept of "Cognitive Computing" is a huge change for computer science. The idea is that these new systems will have the ability to learn, to draw conclusions, to improve every day based on the information they have access. It's a huge challenge for science and demand not only great ability, but also the development of highly sophisticated algorithms, related to the field of Artificial Intelligence.

Watson is much more than a computer, it is the centerpiece of an IBM strategy to develop computational systems with cognitive ability. It is also much more than just a computer ready to play games like Jeopardy.

Healthcare and implications of Watson 

One of the main applications for Watson is in medicine. Traditionally, it  was taught unidirectional as well as almost all other sciences. Teachers taught classes for dozens of students who used the library for more information. The practical, day-to-day aggregated experience the doctor was forming over the years. The best doctors were usually those with enough expertise and experience, which was obtained after 30 or 40 years in the profession.

To try to help change this model, IBM has partnered with Sloan-Kettering Cancer Center, a distinguished institution reference in oncology, with about a thousand researchers who diagnoses and treats approximately 125,000 patients per year. According to Dr. Craig Thompson, CEO of the hospital, work began with a specially developed training for a selected group of researchers of the institution. The idea was to develop cognitive skills, fundamental analysis for the proposed work with the Watson project. They attended to  two years of training at the end of which developed a working model, developed a partner of diagnosis and treatment of cancer.

In general, the working model consists of 3 steps:
  1. Get all the information required for each patient. The goal is to have a methodology to collect laboratory data, clinical examination, imaging, etc..
  2. Using the cognitive skills to "connect the dots", seeing connections and information where it was not possible before, among the thousands of available information from each patient and each treatment method available.
  3. Provide a prioritized list of diagnoses and treatment suggestions, 100% individual, making the process unique for each individual.
The above methodology is already revolutionizing the treatment of cancer. Individualized, unique, treatments was unimaginable a few years ago. Just to get an idea twenty years ago there were only 4 possible types of treatment for breast cancer. Currently there are over 800. How to diagnose and treat appropriately with so many choices? How to ensure that a physician alone is able to analyze all patient information, to know all the different methods of treatment and prescribing correct? With the help of Watson, as a decision support tool, we are moving in this direction.

Of course we still have a long road ahead of us but you can already see significant progress in medicine and other sciences. We will face many challenges, but the rewards, to make a better world, is worth the whole journey.

In the early 90s I studied at Coppe, Coordination of Graduate Programs in Engineering, Federal University of Rio de Janeiro. Followed my specialization in Artificial Intelligence and performed "sophisticated algorithms" on microcomputers even without a hard drive. It was interesting, and we could imagine some pretty cool stuff. Imagine with Watson ...

Want to know more about Watson? Watch the video below and understand the projects and impacts that it can bring to our world.

Nenhum comentário:

Postar um comentário