An approach to predict component failures of imaging system with automated model building process

This session is inspired by a real-time machine learning application that has helped the health facilities to achieve zero downtime due to machine breakdown, and thus, leading to significant savings in terms of time, money and better patient experience.

The session would focus on a machine learning based solution to predict component failures in health care equipment, and thus, the overall learning would be beneficial to data scientists, operational teams and remote monitoring engineers across different businesses.

Key Takeaways

• A detailed approach to work towards a machine learning implementation right from defining the problem statement to productionizing the final solution.
• Identifying the best-fit model by analysing the feature importance and related metrics, thus explaining the results to both data science as well as non-data science audiences.
• Automating the complete process with minimum human intervention to achieve greater overall efficiency and minimize the learning barrier during the process.

————————————————————————————————————————————————————

Bio

Tanmay Gupta – Senior Data Scientist | Philips

Tanmay is Senior Data Science professional with 13 years of experience in life sciences & health care (LSHC) domain and a proven track record of implementing large scale machine learning (ML) algorithms that have provided significant business value to the stakeholders.
Awarded for his work at his present organization, he is skilled in applying machine learning and deep neural techniques to real world health care problems.
He possesses deep knowledge of cloud-based platforms and have designed & implemented scalable ML solutions on commercially available platforms such as AWS & Google Cloud.
He has authored multiple research papers and presented his work in both national and international conferences related to data informatics and health informatics. One of his research work is published as a case study in Harvard Business Journal as well and is primarily used to teach students in the data science domain.