Digital Health Innovations and Acquia spearheaded an event on February 23rd to promote thought leadership on the topic of Machine Learning in healthcare and bring together the San Diego tech and healthcare community.
About Digital Health Innovations (DHI)
Achieve Internet started Digital Health Innovations with the mission to promote collaboration within the San Diego tech and healthcare community.
We aim to bring biotech, life sciences, and mobile health technology companies together by creating events that will enable thought-provoking conversation on the latest innovations that are impacting the healthcare industry.
About the Speaker
Katherine Bailey is the Principal Data Scientist at Acquia and is responsible for the company's strategy around machine learning. She began at Acquia in 2012 as a software engineer on the Site Factory team and then participated in the launch of Acquia Lift which exposed her to the world of data science and machine learning. With her extensive knowledge about software engineering and her many certifications in machine learning, she is able to provide insights on this incredibly interesting topic as a whole and how it is changing our world.
Exploring the Topic: Machine Learning and Healthcare
What is it?
The official definition of machine learning: "a method of data analysis that automates analytical model building. Using algorithms that iteratively learn from data, machine learning allows computers to find hidden insights without being explicitly programmed where to look."
When broken down machine learning can be understood as a form of artificial intelligence that is able to:
- Learn from data
- Recognize patterns
- Apply that intelligence to new data without interference from a human
Why do we need it?
The emergence of touch points in our digital world include the Internet of Things, the multitude of devices consumers own, and mobile applications has led to an enormous growth of data. All of these touch points constantly track and store valuable data about consumers, however, in order for this data to be useful, machine learning must be utilized.
To put this data growth into perspective, Katherine shared that in 2015 the amount of data accumulated in two days matches the amount produced from all years prior to 2013. Data growth is expected to be 350% higher in 2019 than in 2015.
The need for intelligent systems to analyze and learn from the copious amount of data is imperative if we want to make smarter business decisions.
Machine learning creates many benefits for companies looking to improve their customer relations and business practices. ML allows companies to:
The Future of Machine Learning
The future of machine learning rides on the idea that we can create smarter content through auto-tagging, content discovery, and recommendations tailored to individuals. Two key components that will stand out in the future of machine learning include predictive scoring and automation.
Predictive Scoring improves customer profiles and relations by predicting who is a qualified lead and who is not based on various inputs such as social information, demographics, behavioral data, and more.
Automation creates insight on which content to suggest through the extraction of correlations between articles or blogs as well as enable targeted ads to the right customer.
Katherine discussed the potential ethical dangers that can arise due to the utilization of machine learning in our world today. While the idea may be exciting to divulge into, it is important to be mindful of the bias and challenges of this artificial intelligence tool and how it can have harmful effects on our society.
A few ethical implications:
- Machines are taught biased understandings based on the words that humans teach them
- Developers working on self-driving cars are forced to decide whether to program the car to save a passenger or a pedestrian crossing the street
- Machines are only as ethical as the humans who program them; Humans are naturally flawed.
- Companies developing software that can determine stereotypes based solely on image recognition of features - including who is classified as a terrorist based on facial features
The most pressing issue we face in machine learning today is the improvement of natural language processing in order to make increasingly accurate and precise assumptions and predictions.
Machine Learning in Healthcare
Machine learning is making an enormous impact on healthcare today. By utilizing artificial intelligence, researchers are able to better understand cancer through analysis of genome data. Doctors can predict the most effective treatment and diagnosis for specific cancer types based on what the ML program has learned from similar cases. This creates highly personalized care and improved treatment of patients.
When IBM decided to launch the cognitive technology for machine learning, IBM Watson, they chose to focus on the healthcare industry. Their initative for genomics can be described as "Bringing the promise of precision medicine to more cancer patients, Watson can interpret genetic testing results faster and with greater accuracy than manual efforts."
Path Ai is a startup founded by researchers who won a competition on diagnosing breast cancer through reducing error rate in diagnosis by 85%. The researchers went on to create Path AI which develops artificial intelligence applications to improve diagnostics.
i2B2 is a NIH-funded National Center for Biomedical Computing based at Partners Healthcare System. They are developing a scalable informatics framework that would enable clinical data researchers to use the existing clinical data for discovery research. i2B2 offers a web-based tool for researchers to ask questions like "Of the last 10,000 women ages 40-50 years old diagnosed with cancer, what were the treatments offered and their outcomes?"
Researchers at Stanford are currently training neural networks to distinguish between benign, malignant, and non-neoplastic lesions for the dermatology industry. Stanford is also using data to predict when an intervention is needed in the ICU to ensure it is done on time.
Collaboration is Key
In order to progress within the healthcare industry, collaboration with the tech community is vital. This event allowed both sides of the spectrum to contribute different perspectives on the topic of machine learning and bring intelligent voices into one room to share opinions. Questions concerning the reliance on machine learning in healthcare and its potential for error sparked conversation on whether or not physicians should trust machine learning on making life or death decisions or simply rely on their own education and expertise.
The debate continued on whether or not companies should invest more resources in the technology behind machine learning systems or in the acquisition and collection of big databases. Katherine's answered without hesitation that the investment in data will always be more important than the technology. She continued to argue that the right questions backed with meaningful data pave the way for accurate discoveries, insights, and predictions from machine learning technology.