NVIDIA Clara: A successful healthcare application framework

When oneself is in a hospital, they may hear many appreciative remarks, such as “Thank you, Doctor” or “Thank you, Nurse”, & sometimes expressions of gratitude for the technology available. Although there is a substantial amount of appreciation for medical equipment such as MRI scanners, adjustable hospital beds & surgical robots, we often do not realise the formidable software behind all these workings. Processing, acquiring & storing our data safely, accurately & efficiently… is what NVIDIA Clara excels at.

(1)NVIDIA Clara is versatile in creating new computing paradigms to meet healthcare demands. There are four main areas in which Clara can help institutions harness Artificial Intelligence (AI):

·       Drug Discovery

·       Genomics

·       Medical Imaging

·       Smart Hospitals & Medical Instruments

(4)Before we explore the remarkable elements of NVIDIA Clara, I believe it is important to give context on NVIDIA as a company first. NVIDIA is a leader in Graphics Processing Unit (GPU) computing, creating 3D graphics for scientists, designers, artists, gamers & computer users. There may even be an NVIDIA GPU in your computer that you may be using to view my essay right now! From being utilised in the world’s largest gaming platform GeForce, to powering the world’s fastest supercomputers, NVIDIA is the pioneer in GPU computing.  

In the advent of AI, big data, & deep machine learning, many industries and professionals will need to use frameworks like Clara to utilise data to its maximum potential. Thus, let us delve into the unique versions of Clara and their respective functions, that facilitates a variety of procedures done daily in the healthcare sector.

Drug Discovery

(2)NVIDIA Clara Discovery is a collection of frameworks, applications & AI models that use High Powered Computing (HPC) to accelerate drug development processes bringing medicine to the market faster.

(3)Another notable feature of Clara is that is compatible with various software. RELION (REgularized LIkelihood OptimizatioN) is an important tool for studying living cells through electron cryo-microscopy (Cryo-EM). Clara Discovery provides refinement of single or multiple 3D constructions made by this application, hence giving more accurate models.

As well as Protein Modelling, Clara Discovery can also help with Molecular Simulations, mainly of biomolecular systems like proteins, lipids & nucleic acids. Features include visualisation, colouring & a variety of representations of structures.

(5)A client of Clara Discovery is GlaxoSmithKline (GSK), who are utilising the framework for genetics & genomics, modelling of nucleic acids, to improve the development & design of medicines & vaccines.

We’re building new algorithms and approaches in addition to bringing together the best minds at the intersection of medicine, genetics and artificial intelligence in the U.K.’s rich ecosystem. This new partnership with NVIDIA will also contribute additional computational power and state-of-the-art AI technology
— Dr Kim Branson, senior vice president and global head of AI & Machine Learning at GSK
1- GENOMICS.png

The Clara Genomics Technology Stack. Credits: NVIDIA

(6)There is also another part of Parabricks…Clara Parabricks Pipeline which is based on Broad Institute’s Genome Analysis Toolkit (GATK), another genomic analysis toolkit focused on variant discovery. Its functions are to call variants from germline, somatic & RNA sequences using Deep Neural Networks (DNN), and to optimize acceleration, accuracy, and scalability of GPU computing. A prominent feature of this part of the toolkit is that it is fully configurable from the steps, parameters settings, and versions of the pipeline to run, which expands the horizons of what researchers and scientists can do to collected data, truly accelerating the genomics.

(7)Application of Parabricks has been used by the Translational Genomics (TGen) Research Institute. We all are aware that humans are unique on the genomic level, hence having a personalized approach to treatment can be a significant step forward in medicine. They are using genomes of children with rare neurological disorders to look at their genetic fingerprints to decide on treatment. Both researchers & physicians are utilizing this toolkit, as it accelerates genome sequencing of both healthy and diseased cells to assess which form of therapy is the most effective for the individual patient. Sequencing and analysis at such fast rates can allow personalized treatment to be an integrated service in medical workplaces in the close future.

Running Parabricks on NVIDIA GPUs significantly speeds up sequence analysis
— Glen Otero, Ph.D., Vice President of Scientific Computing, Translational Genomics Research
2-PARABRICKS.png

A high-level workflow, transitioning from a sample to final analytic output. Credit: NVIDIA

Apart from clinical medicine there are many other fields Clara Parabricks can facilitate genomics in, as shown above. Its extraordinary to have such advance AI tools that crossover different regions of medicine providing a multitude of solutions. I believe it is important to understand the complex nature of such cognitive computing that makes a framework effective in different disciplines and efficient at its respective functions.

Medical Imaging

Yet another exceptional framework of NVIDIA Clara which is appropriately named NVIDIA Clara Imaging, which is made for the development & deployment of AI in medical imaging. With robust AI models and scalable end-to-end software framework, it is used widely in Magnetic Resonance Imaging (MRI), Mammography, Computed Tomography (CT) and more.

3- CLARA TRAIN.png

Clara Train collaborate and build AI models Credit: NVIDIA

(9)Like Clara Parabricks there are two parts to this application framework: Clara Train Application Framework and Clara Deploy Application Framework. Clara Train includes APIs which allow AI-assisted annotation to images. Using privacy- preserved federated learning allows multiple institutions to collaborate and share AI models without sharing private data. Clara Deploy, as suggested, deploys scalable multi-AI imaging workloads that include DICOM for communication interfaces with hospital systems infrastructure.

An AI model is only as good as the data it processes, hence federated learning is truly valuable as it lets models gain experience with real data whilst respecting the privacy and confidentiality of patients. With limited synthetic data, the framework would not be as robust or provide as many advantages to researchers and physicians as there may be a large sum of false positives due to bias inputs. This also reduces how viable this idea may be when applied to medical settings, hence this is an essential aspect of Clara Imaging.

4 - FEDERATED LEARNING.png

A centralized-server approach to federated learning Credits: NVIDIA

Simply Federated Learning itself can transform healthcare, allowing institutions to work together safely. The NHS can benefit from this as they can ensure they are providing the same high-quality service throughout all their workplaces.

(8)Enabling large volumes of diverse data, clinicians would be able to access a range of data for a wide demographic of patients from different areas. Hence, they can direct their work to fulfill healthcare demands and later can even feedback to the system on its performance.


Smart Hospitals & Medical Instruments

(14)Let us investigate this final and most interactive form of AI NVIDIA has to offer for healthcare. NVIDIA GPUs are paired with sensors, gauges, probes, and deep learning to offer an intelligent tool kit for hospitals. With federated learning, speech recognition, electronic medical records, conversational AI, and more is made possible to optimize care and treatment in a healthcare setting.


(10) Keeping us safe in medical areas is NVIDIA Clara Guardian, the AI framework for Smart Hospitals. An extra set of ears and eyes in the form of sensors can help in ensuring high-quality protection. The COVID- 19 pandemic has emphasized how critical it is to have hygiene etiquettes, awareness of your body, and others, hence Clara Guardian with a whole host of sensitive hardware can perform temperature screening, check if Personal Protective Equipment (PPE) is appropriate, and recognize social distancing. This is only in terms of safety during an outbreak, however, there is a range of functions Clara Guardian can perform.

To see the benefits in action, (11)DARVIS (Data Analytics Real-World Visual Information System), uses Clara Guardian to ensure repetitive parts of healthcare professionals’ jobs are done correctly and are not neglected. This can save time, money, and effort in the long run.

It’s all about compliance. It’s not about surveilling workers but giving them feedback where they could harm themselves. It’s for both worker protection and patient security.
— Jan-Philipp Mohr, Co-founder and CEO of the DARVIS
6 - DARVIS.png

Workings of DARVIS to feedback on physicians Credit: DARVIS

Innovations that are developing like DARVIS, are ones of (12) temperature detection in not only hospitals, but in workplaces, schools, and stores to ensure large scale safety. As NVIDIAs frameworks optimize and accelerate software, temperature readings may be over 100 in a minute! (13) Contact surfaces can also be monitored by AI models in the same way humans are, limiting the number of contacts in a healthcare setting will inevitably lead to a lower rate of transmission of diseases. We must recognize that seemingly insignificant actions may over time lead to consequences, thus by taking responsibility to ensure that we follow the protocol and develop beneficial habits we can save money, effort, time, and lives.

To conclude, it is evident that NVIDIA Clara is an all-rounder when facilitating the healthcare environment. Whether monitoring people, scanning patients, analyzing our DNA, or modeling medicine. Clients of NVIDIA Clara appreciate the optimization and acceleration the framework has given to their product by using AI and deep learning, and the development of such innovations brings us closer to a highly advanced future for medicine and a healthier future for people.


As always thank you for reading and stay safe!

 

 

 

 

References

(1)  NVIDIA – Healthcare & Life Sciences Homepage (n.d) [ Accessed 20:23 17/02/2021] https://www.nvidia.com/en-gb/industries/healthcare-life-sciences/

(2)  NVIDIA – NVIDIA Clara Discovery (n.d)  [ Accessed 11:13 18/02/2021] https://www.nvidia.com/en-gb/healthcare/clara-discovery/

(3)  NVIDIA – Catalog Clara Discovery(n.d)  [ Accessed 11:30 18/02/2021] https://ngc.nvidia.com/catalog/collections/nvidia:claradiscovery

(4)  NVIDIA – About Us (n.d) [ Accessed 12:03 18/02/2021]                  https://www.nvidia.com/en-gb/about-nvidia/ai-computing/

(5)  NVIDIA- NVIDIA Announces Partnership with GSK’s AI-Powered Lab for Discovery of Medicines and Vaccines (Monday, October 5, 2020) – [ Accessed 13:01 18/02/2021] https://nvidianews.nvidia.com/news/nvidia-announces-partnership-with-gsks-ai-powered-lab-for-discovery-of-medicines-and-vaccines

(6)  NVIDIA- Clara Parabricks (n.d)  [ Accessed 13:27 18/02/2021] https://developer.nvidia.com/clara-parabricks

(7)  [Success Story] TGEN uses NVIDIA Clara Parabricks Pipelines for Personalized Patient Treatments (n.d) [ Accessed 14:00 18/02/2021]   https://resources.nvidia.com/en-us-pm-genomics/tgen-success-story

(8)  NVIDIA- Nicola Rieke; What is Federated Learning? (October 13, 2019) – [ Accessed 14:47 18/02/2021] https://blogs.nvidia.com/blog/2019/10/13/what-is-federated-learning/

(9)  NVIDIA- Clara Imaging (n.d) – [ Accessed 15:21 18/02/2021]      https://www.nvidia.com/engb/healthcare/clara-imaging/

(10) NVIDIA – NVIDIA Clara Guardian (n.d) – [ Accessed 16:21 18/02/2021]

(11) NVIDIA – Angie Lee; Smart Hospitals: DARVIS (July 17, 2020) [ Accessed 16:57 18/02/2021] https://blogs.nvidia.com/blog/2020/07/17/smart-hospitals-darvis/

(12) NVIDIA – Renee Yao; Taking the Heat Off (July 31, 2020) [ Accessed 17:34 18/02/2021] https://blogs.nvidia.com/blog/2020/07/31/ai-temperature-screening-covid/

(13) NVIDIA – Renee Yao; Start-up’s AI Platform Allows Contact-Free Hospital Interactions (Sept 3, 2020) [ Accessed 18:00 18/02/2021] https://blogs.nvidia.com/blog/2020/09/03/ouva-contact-free-hospitals/

(14) NVIDIA- Smart Hospitals (n.d) [ Accessed 18:32 18/02/2021] https://www.nvidia.com/en-gb/industries/healthcare-life-sciences/smart-hospitals/