The Project

  • Virtual sensors based on Machine Learning for intelligent diagnosis

Advanced support system for endurance training based on a Virtual Lactate Sensor implemented by Artificial Intelligence techniques (IG-2015/0000062)


During the last years, it is clear the growing importance of endurance sports, especially middle- and long-distance races, included or not within the also very popular endurance triathlon. Currently, these athletes support the workouts on a smart watch or heart rate monitor as well as software applications for analysis and statistics. These software applications make possible to analyze a posteriori data itineraries, to make comparisons, to obtain statistical values, ... facilitating the realization of comprehensive and systematic analysis of the progress of the athlete throughout the season and, therefore, supporting the establishment of training guidelines. One of the most important aspects to consider when planning the season's training is known as "lactate threshold". Nowadays it is fully demonstrated that the lactate threshold, ie, the exercise intensity at which the concentration of blood lactate begins to significantly increase compared to the values at resting, it is more decisive for sports performance than other variables such as the maximum consumption of oxygen or the running economy. However, to measure the concentration of lactate is usual to perform a specific test that requires a blood sample and lactate meter. This test is usually performed in a specialized center and involves a high cost. Thus, nowadays an amateur athlete can not obtain on his/her own the lactate threshold without resorting to a specialized center and/or using specialized and expensive devices. Therefore, it is proposed to research on the design and development of a Virtual Lactate Sensor to estimate the lactate curve and threshold, without adding to the athlete's body any additional element, nor blood sampling or using a lactate meter.


Amongst the achievements of the project, we can highlight the following:

  • An extensive database consisted of 210 lactate testing experiments.
  • Design of a first version of artificial Neural Network system that predicts the lactate threshold from easilly measurable parameters and variables.