Data analysis drives breakthroughs in medicine

Hospitals are usually the busiest places, but the neonatal intensive care Unit (NICU) of the Toronto Children's Hospital in Canada can handle more than ever before. This is thanks to the help of new technology partners. The hospital is using analytics to more accurately predict which preterm infants are at greater risk of disease and infection.

The children's Hospital in Toronto is conducting research on monitoring the temperature, heart rate, oxygen saturation and blood pressure levels in preterm infants by collecting flow data from electronic devices that monitor preterm infants.

At present, from insurance companies, hospitals to service providers, health care providers are actively seeking to adopt advanced technology to help better care for patients. In many cases, they have become more forward-looking and personalized than ever, and have been able to save more medical costs for patients.

"Technology can make a process change our lives," said Susan Feldman, vice president of marketing research firm IDC. The healthcare industry is beginning to understand that we need to make more decisions based on evidence. In this way, we can make the process more reasonable, more accurate identification of information, but also in an operational manner to provide them to decision makers. ”

It is not just the healthcare industry that benefits from predictive analysis, but other industries that are plagued by growing numbers of data. This technology is of great importance in the fields of finance, emergency response, entertainment and law.

"Today, society can do more than ever to support decisions in ways that were not possible in the past," said Dr Harlan Krumholz, a physician and professor of medicine at Yale University, who worked with large hospital alliances to study large data. ”

He warns that meticulous physicians should be responsible for ensuring the correctness of the detailed findings. "These results can raise medical standards and must be closely watched," he said. The question is, what do people do with the information they get from these systems?

"If the inferences about this information are incorrect, then these (incorrect) inferences will lead to wrong decisions, which are clearly not of any value and, conversely, harmful," Krumholz said. ”

Data reduction for Sick children

At present, Toronto Children's Hospital is using IBM's Infosphere streams large data analysis platform to process data in real time. Infosphere streams can correlate and analyze thousands of live data sources. The Ontario Polytechnic University (UOIT) is also using the software to collect streaming data from electronic devices that monitor preterm infants.

Dr. Carolyn McGregor, chief scientist of Health Informatics at the University of Ontario, said the technology enabled the University of Ontario to find out the data and analyse them, such as revealing precursors of sepsis and the many conditions before these problems occurred.

The test has been synchronized with clinical applications, and doctors and scientists are able to compare these two methods, and they are doing it now. A day of valuable data is copied and sent back to the Ontario polytechnic for off-line analysis.

Professor McGregor, associate dean of the Ontario Polytechnic University, said the platform known as Artemis was also referred to as a "data Baby" and was included in a set of clinical rules as an analytic layer to help predict the first time. "At present, the medical equipment on the bedside can provide a lot of information," she said. These devices provide relevant information at very high frequencies, and people have to be able to analyze them, and these results are constantly changing. ”

The final results are not yet available, but they are expected to be peer-reviewed at some point at the end of April and will be released by the end of the year. Preliminary results have shown that Artemis can be a robust approach, McGregor said. She explained that by studying more than 400 patients in three places, they had collected the equivalent of 40 patient data.

Although some traditional indicators, such as body temperature, are available to medical staff at the beginning of the infection, Artemis can provide "a richer environment" to analyse the different symptoms of a baby's condition that may develop at most, McGregor said.

Marty Kohn, IBM's chief medical scientist, said: "The system is designed to predict the onset of sepsis 24 hours ahead of the onset of significant clinical symptoms." The system uses structured data to look for models that allow hospitals to predict the start of a serious disease based on clinical observations. In a similar case, if you can intervene one hours in advance, you can significantly improve your treatment. ”

McGregor said: "In addition to saving the lives of infants, the system can also shorten the average length of hospital stay in infants." ”

McGregor explained that Ontario Polytechnic and hospital staff also plan to use the data to do more clinical research "to find out more about hidden information from the heart rate and other data, and to discover something else from the observation of physiology." She said that by using the new algorithm, Artemis would be able to tell clinicians more accurately about the possible infection of behavioral changes in their infants.

Handling Data floods

Being able to understand unstructured information and predict possible diseases, so that doctors can better symptomatic treatment is the purpose of IBM's tools for developing Watson technology. Doctors are always facing a lot of information, especially thousands of new articles appearing in medical journals every month. Coupled with laboratory results, medical systems and social networking sites for professionals, the data will come at a very fast pace.

"As you get more information, it is impossible for individuals to manage a physical library or to collect electronic collections unless you create a very intelligent way to access it," says Feldman of IDC. Language is so rich, so diverse, that the same concept has many ways of expressing it, so you actually need some technology to understand all the ways we understand things. ”

Tools that can use analytics to organize multiple sources of information are helping industries such as healthcare get a new level of artificial intelligence, observers say. IDC predicts that the entire large data and information technology and services market will grow from $3.2 billion trillion in 2010 to $16.9 billion in 2015.

"In addition to being able to make assumptions through their file-mining and analytics capabilities, Watson has such a breakthrough because it can understand problems and try to translate them," Feldman said. She says IBM built Watson to deal with "deep questions and answers."

"It's a whole new solution or technology that tries to understand what people are looking for," she says. "Technology can understand how an entity affects other entities." In the medical field, for example, computers can analyze how drugs affect specific groups of people with particular types of disease.

This solution works in a completely different way than the search engine works. Search engines try to answer questions that use several keyword inputs. "Then you will get some Web sites or Web pages to see if they are connected to problems in your mind," Kohn said. Then you need to spend a lot of time, which is a very inefficient procedure at best. ”

Instead, by using a series of massively parallel "probabilistic algorithms" to analyze the given information, Watson is able to understand the questions submitted in natural language, read the literature, and then extract and feed back the best advice on the topic, Kohn said.

(Responsible editor: The good of the Legacy)