AgeHack - the first online hackathon to extend life on the MLBootCamp platform

Today, June 15, the ML Boot Camp championship is launched, dedicated to the problems of health and human longevity. The championship was organized by us together with Insilico Medicine in collaboration with the Republican e-health center under the Ministry of Health of the Republic of Kazakhstan. About why this is not a very usual competition for us - under the cut.

This is the third championship this year, but it is very special for so many reasons. Firstly, this is the first championship that we are not holding alone - the competition is held in close partnership with Insilico Medicine and the Republican Center for Electronic Health under the Ministry of Health of the Republic of Kazakhstan.

Secondly, the subject matter of the competition is unusual and very close to the actual problems of humanity as a whole. This is medicine. This time, programmers will look for a solution to the problem of diagnosing cardiovascular diseases. This is the first championship in Eurasia, in which participants use machine learning technologies to find solutions that ensure health and longevity. It seems to us that for the participants it is a very good chance not only to solve an interesting problem, but it is also possible to contribute to the solution of such important issues as the general health of humanity. Who knows, maybe we are standing at the origins of completely new diagnostic and treatment tools. And this is great!

Thirdly,. No logs, the most real clinical data, meticulously collected in medical institutions. 100 thousand anonymized clinical tests. In addition, the mechanics of solving a problem are slightly different from what we usually do at our championships (see the “Problem” section).

Fourth, prizes. Our regular participants are already aware of the current scheme “top6 prizes and 50 souvenir places”. But there is a little surprise, more on that below.

For the rest, the championship takes place as usual - within one month, from June 15 to July 15, 2017, participants must solve exactly one task.

Task "Availability of CVD"

As part of the competition, participants need to predict the presence of cardiovascular diseases according to the results of a classical medical examination. Dataset is formed from 100,000 real clinical tests, and it uses signs that can be divided into 3 groups:

Objective signs:

1. Age
2. Growth
3. Weight
4. Floor

Measurement results:

1. Upper and lower blood pressure
2. Cholesterol
3. Glucose

Subjective signs:

1. Smoking
2. Alcohol consumption
3. Physical activity

Age is given in days. The values ​​of cholesterol and glucose are presented in one of three classes: normal, higher than normal, significantly higher than normal. The meanings of subjective signs are binary.

All indicators are given at the time of inspection. Now a little unusual.

The last group of signs does not have an unambiguous interpretation and, moreover, is collected from the words of the patient himself. Therefore, for part of the data in the test sample, we specifically excluded these parameters. Participants will have to either predict them independently for some cases, or ignore them altogether.

The data is divided in the ratio of 70/10/20. The training sample consists of 70 thousand results, another ten thousand are public metrics that are available to participants during the competition. The remaining 20 thousand went to a hidden test sample, counting the metrics by which will determine the winners in the final.

Since we are dealing with binary classification, the metric is the logarithmic loss function.

Prizes

The distribution of six prizes this time looks like this:

Top1: MacBook Pro
Top2: NVIDIA 1080ti
Top3: NVIDIA 1060
Top4-5-6: WD My Cloud 6 TB

By tradition, the top 50 participants will get T-shirts with the symbols of the championship.

Participants with the most interesting solutions for the organizers will receive an opportunity for internship or cooperation with the Mail.Ru Group, Insilico Medicine and the Ministry of Health of the Republic of Kazakhstan. In addition, a special prize from the jury is a trip to Astana for a personal meeting with the Minister of Health of the Republic of Kazakhstan.

MLBootCamp Community

With each new competition, our Telegram community grows larger. Professionals share experiences, beginners learn. The doors are open to everyone, so join us.

check in

The championship starts today, at 18:00 Moscow time. Register here . Good luck!