Search
  • Barry Holleman

The tales of dirty data – learnings for healthcare AI

Dirty data can easily derail any big data analytics project, especially when bringing together several data sources that may record clinical or operational elements in slightly different formats.  Several data conventions in health care hinder the widespread use of data analytics. Currently, health care data are split among different entities and have different formats such that building an insightful, granular database is next to impossible.

One of the most hyped applications of big data in epidemiology, Google Flu Trends, turned out to underperform far more basic models, despite analyzing far more data, because its analysts were extrapolating from the behavior of Google users—an unrepresentative group of people. The experience illustrated that the success of data analytics in health care is dependent upon the availability and utilization of quality data.


Data cleaning ensures that datasets are accurate, correct, consistent, relevant, and not corrupted in any way. While most data cleaning processes are still performed manually, automated tools that use logic rules to compare, contrast, and correct large datasets can dramatically reduce this effort.  These tools are now more sophisticated and precise as machine learning techniques have demonstrated their effectiveniss, reducing the time and expense required to ensure high levels of accuracy and integrity in healthcare data warehouses.


When it comes to machine learning, the applications are data hungry. The more high-quality labeled data a developer feeds an AI model, the more accurate its inferences. Creating robust datasets remains an obstacle for data scientists and developers building machine learning models.


Understanding, designing and executing a data labeling workflow has often proven to be a time-consuming exercise. With the advancements in AI and expertise in healthcare specific data labeling workflows, the required effort has now significantly decreased and carries promise for the application of AI in general and the healthcare setting specifically.


How? The higher the data quality, the less data needed to achieve accurate results. A machine learning model can produce the same results after training on a million images with low-accuracy labels, or just 100,000 images with high-accuracy labels.


Healthcare specific data management tools and healthcare specific labeling processes are important components of the MUUTAA platform in order to deliver workable annotated data sets that increase the efficiency of training AI models and improve the accuracy of the models in the shortest possible period of time. Embedded in the MUUTAA framework, meeting the industry’s privacy and security requirements, these tools can be deployed in a cloud or on premise setting.

54 views

Site Navigation

Useful Links 

More MUUTAA

Contact MUUTAA

linkedin-logo_edited_edited.png
  • Twitter

©2020 MUUTAA. Created by DukeVision. Legal Notice.