Connect with us


EXCLUSIVE: GSK: Key factors driving digital transformation



The healthcare industry is entering the era of digital innovation, and we already see the possibilities the digital transformation in healthcare can offer.

We asked Fausto Artico, Global R&D Tech Head and Director of Innovation and Data Science at GlaxoSmithKline (GSK), the 6th largest Pharma company in the world, to share his views on digitalisation trends in healthcare. For the past two years, he and his teams have been working on digitising and innovating the pharma landscape, allowing GSK and its partners to shorten time to market, improve manufacturing, and enhance customer experience.

In your opinion, what are the key factors driving the healthcare digital transformation?

If I had to choose one factor, I would say it is patient experience. Everyone wants to be treated faster, cheaper, and more conveniently. Patients want a better quality of life, especially if they are afflicted by one or more chronic diseases. To improve patient experience, we can use several emerging technologies of which digital transformation is just one. The two most important components characterising and driving digital transformation are digitisation and digitalisation.

Digitisation is the process of converting something to digital form. A huge amount of data today is difficult to use because it is still in documents that need to be scanned and requires digital capabilities to be processed. Think of all the paper documents that even today need to be created to comply with regulations and procedures. The more manual recording and handling of paper documents is involved, the more difficult it is to provide an improved experience to patients.

Digitalisation is the process of transforming existing business procedures and workflows through the use of digital technologies. Thanks to the IoT, we can generate, collect, and input patient data in ways that are much easier to execute. Thanks to Optical Character Recognition technologies, we can transform data otherwise unavailable for digital processing. Thanks to Big Data, we can integrate and process huge and disparate amounts of digital data. Thanks to AI and ML, we can increase the probability of discovering new drugs feeding back effects and results to the early-stage in discovery, optimise manufacturing processes to create products with greater stability and shelf-life, and enhance patients, doctors, and healthcare providers user experience. Thanks to Hyper-Automation, we can become better at tracking and logging activities, correcting errors and continuously reducing human intervention for cumbersome, tedious and error-prone procedures related to data collection, cleaning, linking, contextualisation and harmonisation. Finally, by leveraging DevOps practices, we can greatly accelerate the ability to test drive on a small scale first and deploy on a large scale next what we discovered through data to continuously and frequently improve patients lives.

There is a wide range of digital innovations companies can leverage to become more competitive. What could be a starting point for a company that wants to successfully become digital-forward?

The starting point depends on the digital maturity of the company and its existing infrastructures. Companies need to be smart. They need to know their core capabilities, and continue their digital transformation journey by building new capabilities from their already very well consolidated core competencies.

My advice to companies is to start with something easy, for example creating digital dashboards to increase transparency and visibility of processes. This will allow them to discover where their problems are, usually in the form of inefficiencies, defects, long lead times, variabilities, etc. Сompanies then can use ML algorithms to try to solve them, thus reducing waste (if their competitive advantage is cost), or create new products and services (if their competitive advantage is differentiation) with whatever data they have available in digital form. If the data is deemed to be insufficient, they can proceed with the creation of data pipelines leveraging Big Data, Hyper-Automation and Optical Character Recognition. All this of course makes sense only if the business benefits that will be provided by the solutions outweigh the total investment necessary for their development.

I have seen too many large-scale projects fail just because companies rushed into the digital transformation journey without having built internal capability or gathered external expertise. If you are new to digital transformation, try to avoid big initiatives. The technological know-how for such initiatives is only one of the reasons why big digital transformation initiatives fail. There are also cultural and political aspects that must be considered. Starting small, with something “easy” and “limited” in scope, allows you to de-risk your digital transformation journey on multiple fronts.

What factors do you think healthcare organisations should consider to make the most of their data?

Companies need to prioritise which of their data sources should be made digital. Usually you need much less digital data than what you think but, unfortunately, you often do not know a priori which are the important data sources you will need to solve your problem. Involving Subject Matter Experts in the Use Cases you are developing is very important because they can help you prioritise and minimise the necessary digitisation work you will need to do to solve your Use Cases.

Fausto Artico

As more and more digital data sources become available, companies need to integrate all of them together in a seamless, continuous and incremental way. This is extremely important because the number of connections between single data points grows exponentially as more data sources become available. Many of the most complex problems of today usually have multiple root causes that require large scale discovery activities to be solved. Such activities are possible only if multiple data sources are integrated. For example, think of a supply chain and how something happening during manufacturing could affect patients when they receive a treatment if there isn’t a full end-to-end track and trace monitoring system available for consultation.

You mentioned some technologies that are important in the digital transformation journey. Could you please elaborate on why they are necessary to be successful in the healthcare market?

The Internet of Things is revolutionising the way we create wearable and under-the-skin devices. Such devices are extremely useful to automate the generation, gathering, and processing of digital data. They allow single sign-on authentication for multiple systems, can execute quick periodic tests during clinical trials, enable the monitoring and logging of any kind of activities, etc.  All this is possible thanks to our ability to manufacture cost-effective miniaturised sensors, interconnect them to easily configurable large-scale networks and make it possible for them to exchange and transfer data using low power consumption hardware.

Optical Character Recognition is making it possible to digitise paper documents for processes and procedures that cannot be easily automated. They allow companies to leverage the huge amounts of historical data that today reside in archives where paper documents have been stored for more than 20 years because of compliance and regulatory reasons. Often, the paper documents contain data that cannot be found anywhere else and therefore they can be extremely important for the successful execution and delivery of new capabilities through AI and ML solutions.

Big Data allows companies to automate these processes, determine which data is reliable, link and organise it in different types of data structures to accelerate data retrieval and usage, as well as, for the first time ever, get holistic, real-time, end-to-end views of complex processes. Automation procedures can generate data dictionaries explaining the meaning of different data types to non-expert people coming from disparate, non-technical domains. Visual tools simplify data exploration (e.g hypothesis generation), allow people to easily understand problems and communicate new insights to business stakeholders and senior level people in ways that are easy to understand.

AI and ML can greatly simplify our job of discovery and generate new ways of solving complex problems beyond our conscious ability to interconnect data and discover interesting patterns between networks of data points. As more and more data becomes digitally available and integrated, it will also be important to create easy to explain, understand, and interpret AI and ML. This will be necessary to allow more classical and historical scientific domains to advance our human knowledge with new models (e.g., mechanical ones) that are not purely statistical like the ones characterising the AI and ML domains.

Having the ability to continuously design, implement, test, integrate and deploy into production Big Data capabilities as well as personalising and changing processes through the new results produced by AI and ML models is fundamental in creating and sustaining competitive advantage and leapfrogging other organisations. Software engineering teams can achieve all these goals by creating new products and services handling their whole life cycle from inception to decommission. By minimising the number of handovers among teams, leveraging Agile methodologies of software development, and creating teams owning all the necessary resources, even big and complex companies in very regulated environments can innovate more quickly and reduce lead times to market, increasing their ability to adapt to consumer needs faster, and improving patient experience and quality of life.

There is much debate going on to understand if and how much to invest in robotic automation. What are your views about this? 

It is difficult to think that regulatory organisations will allow companies to completely remove humans from many of the activities they need to execute to be compliant and therefore licensed to provide healthcare products and services. Furthermore, regulatory organisations require easy-to-understand, logical explanations about deviations and cures. Today robots cannot provide such explanations because of their purely statistical nature and are very limited in their ability to execute unclear, ill-defined, unstructured, non-repetitive tasks.

Robots are better than humans at executing tasks that are clear, repetitive, structured and well-defined. There are plenty of such tasks that, if automated, can free humans to invest their time and energy on more value-added activities that robots cannot execute (e.g., research, negotiation, influence, etc). Robots are also better than humans at organising and sifting through huge amounts of data, finding new hidden patterns, anomalies or strange behaviours that seem statistically significant. Additionally, robots are invaluable helpers for many activities humans need to execute manually, or for validating hunches humans may have but would find difficult to test considering our limited conscious capacity to process information.

Humans are better than robots at generating new hypotheses, exploring opportunities in non-conventional ways, working with teams of people, etc. We have a much better holistic view of what is happening than robots. Existing hardware technology does not allow robots to understand all the possible nuances of a situation like we humans do. More importantly, robots do not have emotions or a conscience, therefore making it impossible for them to be aware of and understand ethical and moral issues that need to be considered in many situations. Humans will probably remain the final decision makers for all the important decisions, even if they leverage robots.

Both robots and humans are necessary and fundamental to the success of the digital revolution. Companies should architect their future digital systems in ways that allow them to leverage the best of both worlds, creating contexts and positions in such systems where robots and humans can use their strengths but work in symbiosis to provide much faster and better healthcare to patients in need.

How do you see the future of digitalisation in the healthcare industry in 5-10 years?

We will see more of what is already going on. This means more interconnected devices to further accelerate the automatic generation and collection of data before processing it with Big Data methodologies. We will see more interpretable ML models because, for such an important sector as healthcare, we are not only concerned about the ability of the models to accurately predict outcomes and validate their success rates, but we also want to know why phenomena happen to be able to develop better solutions that would be easier to apply to wider stratums of the population. Automating all digital activities through Hyper-Automation and DevOps practices will further increase the speed with which companies can remain competitive in the market and personalise care and user experience for patients. However, I think there will also be some new interesting technologies and methodologies that will reach the market.

First would be miniaturised hardware devices characterised by biological components. Such devices will enable activities and interventions today believed impossible. Their processing and “reasoning” capability, their ways of configuring and interfacing with other biological components (e.g., lipidic membranes to enter into cells) and the multitude of ways they could interact with the human body are simply incredible. The simplest composition of such devices would be pure standard hardware on the nanoscale level. The most complex compositions would be characterised by process capabilities that would involve enzymes or smaller biological components leveraged as “programming language” and “computational units” to use to accelerate or modify biological processes.

Secondly, we’ll see enhanced cognitive capabilities allowing machines to behave more like humans. This will enable more user-friendly interactions between machines and people, allowing healthcare organisations to simplify and deliver services that are currently impossible to scale due to the limited number of available doctors. Routine procedures, basic prescriptions and diagnoses could be executed through chat boxes leveraging cognitive technologies (e.g. automatic language translation) in ways that will allow the delivery of products and services even to very remote, undeveloped, emerging areas and markets located anywhere on the planet. Expertise developed in industrialised countries will be easily transferred everywhere and a more human touch will be introduced into the technologies to make products and services more usable by people with no knowledge or expertise in software systems.

If you’d like to learn more about the major digital trends in the healthcare industry and meet a lively community of professionals, join Fausto Artico at the Healthcare Automation & Digitalisation Congress 2022, where he will share more of his insights as a speaker at Session 1: You are Data.

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories