The latest offering on healthcare from Amazon was introduced in the form of Amazon HealthLake, which was announced at the on-going AWS re:Invent 2020 event. Amazon HealthLake enables healthcare organisations to store, transform, and analyse large life science data, to the tune of petabytes, in AWS. It can understand and extract medical information that includes rules, procedures, and diagnoses in real-time. Amazon HealthLake is a HIPAA (Health Insurance Portability and Accountability Act)-eligible service for healthcare providers, insurance companies, and pharmaceuticals.
While introducing HealthLake at re:Invent 2020, Swami Sivasubramaniam, the vice president of Amazon Machine Learning for AWS said that while there has been an ‘explosion’ of digitised health data with the introduction of e-medical records, the amount of valuable information that could be derived out of this data hasn’t been able to keep pace.
He then added, “With Amazon HealthLake, healthcare organisations can reduce the time it takes to transform health data in the cloud from weeks to minutes so that it can be analysed securely, even at petabyte scale.”
Why Conventional Methods Fail
A lot of sectors are now applying machine learning to their data to derive maximum value information to apply in their services and processes, and the healthcare industry is no different. This industry requires careful information extraction from important patient data. This requires the application of analytics and machine learning, which in turn improves the care provided, helps in analysis of population health trends, and supports greater operational efficiency.
It is much easier to state in words, but the fact remains that clinical data is highly complex. What makes things more difficult is that it is often siloed, incompatible, stored across different on-premise locations. Apart from gathering these critical pieces of information in the FIHR (Fast Healthcare Interoperability Resources) format, which is the standard that describes data formats for exchanging electronic health records, one also needs to tag, index, and structure data. Only after these processes, data is deemed fit for the query.
As per current trends, healthcare organisations have rule-based tools for process automation to transform unstructured data such as medical histories, physician notes, and imaging reports. Since these are not generalised tools, they break down when applied to other systems or when there is variation in spelling, and have typos or grammatical errors.
Another popular tool used by healthcare institutions is optical character recognition (OCR) software to process data. However, the problem with OCR is that it is not 100% effective, and as technicians have to manually enter data, the process gives scope for error and adds to the digitisation process.
The cost and the complexity of the whole process of analysing and discovering trends to make precise predictions prohibits organisations to fully utilise the potential of data.
How Can HealthLake Help?
Amazon’s new offering brings together all the data and helps organisations to ‘make sense’ of it and make more precise predictions about patient health.
HealthLake stores, tags, indexes, standardise queries data at a petabyte-scale in the cloud. It further allows organisations to easily copy data from on-premises systems to a secure data lake in the cloud and normalise every record across disparate formats.
Amazon HealthLake uses machine learning trained systems to understand medical terminology to identify, tag, and enhance data with standardised labels such as — medications, conditions, diagnoses, and processes. This helps in avoiding potential error that is observed with conventional tools.
Amazon also suggests that when used with other AWS analytics and machine learning services like Amazon QuickSight and SageMaker, technicians will also be able to get interactive dashboards and will be equipped to build and deploy custom ML models.
The launch of HealthLake comes amidst Amazon’s aggressive efforts to charter healthcare AI domain. It may be noted that just last year, the company had launched Transcribe Medical to transcribe medical speech for clinical staff of primary care. A year before that, Amazon had made three HIPAA eligible offerings — Transcribem Translate, and Comprehend.