Blautic artificial intelligence
Our hardware and software development activity for monitoring and data collection has invariably been associated with data analysis collected by our systems. Initially, in custom developments for research projects and ad-hoc solutions for our clients and, more recently, as specific implementations for our products
Blautic artificial intelligence
Our hardware and software development activity for monitoring and data collection has invariably been associated with data analysis collected by our systems. Initially, in custom developments for research projects and ad-hoc solutions for our clients and, more recently, as specific implementations for our products
Custom AI Models
At Blautic, we develop recognition models for many of the signals generated by our devices (ECG, EMG or accelerometers). These artificial intelligence models allow us to real-time check the signal quality and provide our clients with high-value information. We custom train machine learning models with direct or elaborate patterns capable of recognizing, in real-time, any signal that exhibits a pattern of operation
ML Models with Direct Data
Based on data obtained directly from linked sensor devices
ML Models with Elaborate Data
Based on graphs extracted from data such as Lorentz diagrams generated from RR intervals of the ECG signal
Accessible AI Training
At Blautic, we have made a significant effort to make the training of our artificial intelligence systems accessible through our applications. Currently, our AI models can be trained by our clients quickly and easily through our applications
AI Training Process with Blautic Apps
1
Data Collection
Our apps have step-by-step guided processes for the proper collection of data that will be used later for model development. Active aids included in the process include time guidelines and real-time data and activity visualization
2
Validation
Once the recordings are registered, our apps guide users through the validation of recordings by playing back/viewing video recordings and multimedia data simultaneously, helping users to clearly distinguish between correct and incorrect recordings
3
Testing
Once our cloud storage systems process the data and generate the models, we can download them and verify their performance. If necessary, we can improve them with new data collection, validation, and data processing
IA Model Market
In addition to allowing coaches and physiotherapists to train exercises, our applications provide a Model Market where they can share their models with other users of the application.
Commitment to Explainability of Our AI
At Blautic, we are committed to addressing the growing concern about opacity and lack of interpretability in AI models based on so-called black boxes. That’s why, whenever possible, we work with XAI (Explainable AI) models to provide understandable and transparent explanations of how our systems make decisions or predictions
Blautic AI Products
Exermeter
An automated exercise quantification system during physical workout sessions. It uses machine learning models based on wearable data from accelerometers, gyroscopes, ECG, and machine learning. Professionals are responsible for training the models and assigning them to users. They receive feedback on sessions performed by users
Ziven Cardio
Applies trained models for recognizing ECG signal quality and classifying Lorentz plots into different types of anomalies
Ziven Active
An application for controlling rehabilitation exercises with motion and muscle activity data. It monitors the completion of repetitions based on schedules assigned by professionals. Users receive online feedback on the correctness of each repetition and explanations regarding what they do not execute correctly
PikkuCam
Allows users to recognize trained exercises through machine learning and create intelligent videos whenever they occur during the activity’s development
