I recently participated in Amazon Web Services’ annual developer conference, re:Invent. In this text, I’ll dig into the insights we’ve gathered regarding AWS and the healthcare and life sciences industries: what AWS is doing and what you should know if you are digitalizing these industries, especially if AWS is part of your technology portfolio or you are figuring out what cloud technologies could offer. For our other insights from re:invent: “It’s all about the money.”

Amazon recently acquired the US-based healthcare service provider One Medical to expand its operations to the healthcare industry, so it’s fairly safe to say that healthcare is firmly part of Amazon’s strategy in the cloud and beyond.

Instead of digging into the possibilities of using generic AWS services, we will focus on industry-specific services and examples, with the exception that one of the cases presented is focused more on the insurance industry but with a strict focus on healthcare insurance.

So what can these services be used for? 

Real-life Examples of What’s Already Possible

Rush University in Chicago has been dealing with the challenge of how the life expectancy of citizens differs drastically in different community areas. The key hypothesis they’ve been working with is that the difference in income levels, and therefore access to high-quality food, correlates with the average life expectancy. Plainly put, in poor community areas, people don’t have enough money for food, which leads to lowered life expectancy. 

By using Amazon HealthLake, they have combined electronic health records (EHR), social determinants of health (SDoH), and cardiometabolic data from disparate systems. They’ve been able to build tools for healthcare professionals that enable them to better care for patients by having a more holistic view of their situation.

Another excellent example of integrating multi-modal health data comes from AWS. They combined Amazon HealthLake and Amazon Omics to showcase how the services can be used in treating lung cancer. Amazon HealthLake Analytics was used to identify a cohort of patients with the same lung cancer diagnosis. Then, Amazon Omics was used to identify patients with a variation of lung cancer susceptible to drugs by searching for specific biomarkers in the sequencing data of their cancers against data from a publicly available database. Finally, Amazon HealthLake Imaging was used to analyze the progression of the cancer from the images taken throughout the treatment.

One more example we would like to highlight by AWS is about intelligent document processing (IDP) of healthcare insurance claims. IDP, in practice, means automation of document processing using AI. Amazon Comprehend Medical was used to identify health data and protected health information (PHI) from medical documents that are part of the claim package, such as doctors’ notes, medical charts, lab reports, and transcripts, so that a claim handler doesn’t need to read through all the documents and interpret the relevant information. 

AWS emphasized that there should always be a human doing the final check on the data, so don’t expect to fully automate the process if you want to avoid legal troubles. Amazon Comprehend Medical can, however, vastly improve the claim-handling process by reducing the manual work required to gather the relevant data from the claims package.

Peeking at the Future: What You Could Do with AWS Tools

To give you a better idea of what you could do with the specific services AWS offers for healthcare and life sciences, we put on our thinking caps and came up with some very raw concepts that we think would be possible with them. I’ve added a short description of the tools mentioned here at the end of this blog.

Intuitive intake

When a patient comes in for a consultation you could replace the laborious intake questionnaires with easier-to-answer and more intuitive alternatives where patients could write free text or even dictate their answers. Services used: Transcribe Medical and Comprehend Medical

Aided consultations

During a consultation, be it remote or at a clinic, the conversation could be transcribed, and turned into structural data. This structural data could then be used to help the clinician to look up relevant data more efficiently or to make their interaction with the EMR much easier and shorter, freeing up their time for the vital part, which is treating the patient. Services used: Transcribe Medical and Comprehend Medical

Patient security 

During a consultation or an operation, the discourse could be transcribed so that it could be more easily analyzed to identify any complications or potential mistakes that might have happened. Services used: Transcribe Medical

Multi-morbidity identification and treatment

By collecting data from disparate systems used at social work, clinic, and hospital surroundings, be it unstructured or structured, to a single place where it can be analyzed to identify patients with multimorbidity and to provide a holistic view of their situation to produce an integrated care plan would enable better care for the patient and ease the burden of the healthcare system. Services used: HealthLake

Care delivery development

By combining data from clinical and hospital systems with patient-reported outcome and experience measures, you could analyze existing care pathways and how effective different variations are to empower the care personnel to proactively develop care delivery to reduce unnecessary care, increase adoption of best practices, and improve outcomes. Services used: HealthLake

Population management

By collecting data from clinical and hospital systems to a single place where it can be analyzed, you can provide population-level insights for client organizations or clinics and hospitals to identify both positive and negative trends that help you guide your efforts to improve well-being and health outcomes. Services used: HealthLake

Prevention

Using both clinical and omics data, you can identify trends and genetic dispositions that need to be addressed when treating patients to provide the best possible care. Services used: HealthLake and Omics

Increased speed of R&D

Reducing the time and effort it takes to set up new data pipelines to analyze omics data when identifying new genetic disorders or dispositions and developing new treatments enables researchers to focus on the critical part of their jobs and provides better and cheaper treatment options for patients. Services used: Omics

Common Goal: Fixing Disparate Systems and Fragmented Data

Here, we’ve highlighted only a few examples of what you can do with what AWS offers for the healthcare and life sciences industries. All the services have one thing in common, and it is data. 

AWS is tackling one of healthcare’s fundamental problems: the inability to provide the best possible care due to disparate systems and fragmented data. 

To do that, AWS has focused on providing ways to store, query, process, and analyze healthcare data from existing core and care delivery systems in its many forms rather than offer new core systems for the industry. Using these services, you can build complementary solutions that can be utilized to improve care delivery on both individual and population levels and increase the speed of innovation in the research and development of new processes, services, and treatments.

The big question is, what could your organization use them for to aid the work of healthcare professionals and provide better care for your patients? 

Our Reaktor Health team would love to discuss and help you figure out how to use these or any other technologies to devise a cloud strategy that enables you to realize your healthcare visions. 

Learn more: Reaktor Health

A short recap of AWS services for healthcare and life sciences

• Amazon Comprehend Medical is a HIPAA-eligible natural language processing (NLP) service designed to identify healthcare data, including ICD-10, SNOMED CT, RxNorm, and PHI. It can be used to, e.g., identify diagnoses and PHI from free-text patient dossier entries to transform the unstructured data to structured data, which can be further utilized for analytics purposes. en-US
• Amazon HealthLake is a HIPAA-eligible data lake developed for healthcare and life sciences use, allowing you to store, query, and analyze all your health data from disparate systems in one place. It can be used to, e.g., combine multi-modal healthcare data to provide individual and population-level insights that enable better care and improved operations. en-US
• Amazon HealthLake Analytics is an additional module in Amazon HealthLake providing the ability to standardize all your structured and unstructured health data from disparate systems into HL7 FHIR format, which you can then query and analyze on individual or population level.
• Amazon HealthLake Imaging is an additional module in Amazon HealthLake rather than a separate service, allowing you to store, access, and analyze Digital Imaging and Communications in Medicine (DICOM) data.
• Amazon Transcribe Medical is a HIPAA-eligible automatic speech recognition (ASR) service, also known as speech-to-text, with support for healthcare terminology. It can be used, e.g., to provide consultation transcripts, reducing the time spent writing patient dossier entries. en-US
• Amazon Omics, which was mentioned before, is a HIPAA-eligible service for storing, querying, processing, and analyzing genomic, transcriptomic, and other omics data in formats such as FASTQ, Binary Alignment/Map (BAM), and Compressed Reference-oriented Alignment Map (CRAM). It can be used, e.g., in pharmaceutical research to increase the speed of innovation or to detect genetic anomalies in patients to provide better care. At the time of the publication of this text, it is important to note that there are some limitations to the availability of these services, both region and locale-wise. Luckily with Amazon Omics, the data is not locale-dependent and is available in the U.S., Europe, and Asia Pacific. Other services, however, like Amazon HealthLake, are only available in U.S. regions, and the only locale currently supported by all the services is en-US.

In summary, while the AWS service offering is growing to cover more use cases in these domains, unfortunately, for those not operating in the US, it can still take considerable time to reap the benefits of this development.