Fiducial localization with sparse annotations

A method of training a machine learning model to produce a heatmap, the method comprising:

• providing an annotated waveform to the machine learning model in which a single period is annotated with one or more fiducials;
• analyzing a first output of the machine learning model generated in response to the machine learning model processing the annotated waveform;
• providing feedback to the machine learning model based on the analysis of the first output;
• providing one or more additional annotated waveforms to the machine learning model, wherein the one or more additional annotated waveforms comprise a single period annotated with a specific fiducial;
• analyzing a second output of the machine learning model generated in response to the machine learning model processing the additional annotated waveform;
• determining an accuracy of the machine learning model by comparing the second output with an expected second output to obtain delta that represents the accuracy of the machine learning model;
• determining that the delta is less than a predetermined accuracy threshold;
• and
• in response to determining that the delta is less than a predetermined accuracy threshold, outputting the machine learning model to a system that maps waveforms to the heatmap.

A system for mapping a waveform to a heatmap, comprising:

• a processor;
• and
• computer memory coupled with the processor, wherein the computer memory comprises data stored thereon that, when executed by the processor, enables the processor to:
• receive a dataset representing waveform data;
• window the dataset such that no more than two periods of the waveform exist within a windowed version of the dataset;
• provide the windowed version of the dataset to a machine learning model;
• receive an output from the machine learning model, wherein the output is generated by the machine learning model in response the machine learning model processing the windowed version of the dataset;
• and
• generate the heatmap with the output received from the machine learning model.