Month: December 2023

Why it’s time for B2B marketers to enter the data mesh

Posted on by Matt Godfrey

B2B marketers love data. Marketing was one of the first business functions to put big bets on analytics and automation, and today, the best B2B marketing campaigns are driven by data. It might not always be complete or accurate, but data helps talented marketers set the general direction of their campaigns and pin their instincts on something tangible.

But what if marketers could easily access trusted data (and lots of it) and use that data to deliver better results?

What if they could uncover new insights hidden in data throughout the business – and use them to create hyper-personalised content and more effective campaigns?

What if they could imagine possible futures for their campaigns and quickly test their hypotheses to see what works?

Well, in a data mesh, they can.

What’s a data mesh? And why should marketers care?

In large, complex organisations with monolithic data architectures, accessing timely, relevant insights can be a laborious process. It relies on specialist data teams to drag insights kicking and screaming out of a central data lake.

The data mesh approach helps overcome these difficulties by decentralising the data architecture and making each domain (marketing, sales, product, etc.) the owner of the data it produces. It’s an approach that’s been growing in popularity over the last few years (which explains why tech consultancies often ask us to write about it) as large enterprises look for ways to reduce organisational and operational complexity.

In a data mesh, the people closest to the data are responsible for managing it and using it to create “data products” that solve their most pressing issues or open up new opportunities.

Federated data ownership removes the operational bottlenecks of centralised structures, so marketers can access and use data how they need to, when they need to. And with data products visible and accessible on a self-service platform, everyone can access products built by other domains and combine them in useful new ways.

New marketing opportunities – and responsibilities – in the data mesh

The data mesh approach empowers marketers to cut out the middleman and start experimenting with their data to find ways to improve content and campaign results. When data users become data owners, the possibilities are limitless.

Marketers who build and own data products can understand their customers and prospects better than ever. They can optimise their campaigns on the fly and conduct low-risk, high-reward experiments with different approaches. They can even begin to create the kind of hyper-personalised content and communications that most marketers can only dream of.

More than most business functions, marketing thrives on data from across the organisation. Insights from sales, service, product, R&D, manufacturing, supply chain, and more can all add valuable context to marketers existing knowledge about their customers.

With a data mesh approach, marketers can easily access data products from other business functions to quickly create new capabilities. For example, combining product and sales data products with a customer-intent data product might help marketers target specific prospects with campaigns that are more likely to land.

But before we get too carried away, it’s important to remember that federated ownership also means federated responsibility. In a data mesh, every domain is a data custodian, so marketing becomes responsible for the governance, compliance, and quality of its data.

Meaningful change takes time

Adopting a data mesh approach requires a fundamental cultural shift; it’s a completely different way of thinking about data and how it’s managed, governed, and used.

This shift in mindset includes a switch to what technologists call “product thinking”, where success is defined by the outcomes products deliver, rather than the outputs of projects. It might also require changes in how teams are structured and how they operate. And it will certainly involve fostering a new culture of cross-functional collaboration, as different business units contribute to combined data products.

It’s not something that happens overnight, and it can take years for large organisations to successfully embed the data mesh approach. But if you’re looking for a long-term, strategic approach to getting more bang for your marketing buck, data mesh could be a conversation worth having with your colleagues in IT and elsewhere in the business.

Stay up to date with what’s next in tech 

If you’d like to keep up with how emerging tech trends can have a big impact on B2B marketing – and get practical advice on other ways to maximise the value of your content – sign up for our newsletter.

Five AI-infused technologies that are transforming healthcare

Posted on by Katy Eddy

A few years ago, I had surgery to fix some mobility problems caused by a form of hip dysplasia.

It took a while to figure my issues out; x-rays, CT scans, even an arthrogram-aided MRI couldn’t get a clear enough picture. Eventually, the diagnostics team used my scans to create a 3D model of my pelvis – a “digital twin” of me that showed my consultant exactly where the problems and damage were.

Two surgeries later, I’ve emerged with a sizeable scar and a real appreciation for how complicated the human body is, and how much technology can do to support healthcare professionals.

These days, I do a lot of healthcare-related writing for various clients, with topics ranging from rural care access to the latest advances in diagnostic AI. And that means I get to see a lot of exciting technologies as they emerge.

Here are five technologies that are making a difference in healthcare – and how they work.

1. Deep learning models for diagnostics

Radiologists work through hundreds of scans each day, identifying and recording their findings. Pair that with highly complex cases, potential interruptions, and additional responsibilities, and there’s an enormous cognitive load to contend with. Over a long shift, the risk of a radiologist missing something on an image grows, even when they’re highly experienced.

So what can AI do to help? Deep learning – a form of AI where a machine learns to complete multiple layers of processing – can support radiologists by offering additional detail in the reading process. These models can be trained to look for specific clinical findings, such as masses, bleeds, and breaks, and flag them to the radiologist for confirmation.

Let’s take chest x-rays, for example. They’re the most widely used imaging test in the world, but there are several layers of bone and organ to contend with, which means findings can get lost in the density. One recent study paired 20 radiologists with a deep learning model, which was trained using more than 800,000 previous chest x-rays. Of 127 clinical findings, the model led to a statistically significant improvement in accuracy for 102 of them.

There’s a worldwide shortage of radiologists, pathologists, and other diagnostics professionals right now, and demand is growing as our global population gets older and sicker – so the support of a deep learning model could make a major difference.

2. Virtual reality simulations for clinical training

All clinical professionals need to learn in a practical environment, but not every patient wants to be the proverbial guinea pig, especially in emergency, sensitive, or high-stress cases. Wouldn’t it be so much easier if we could train clinicians in any situation on demand, to make sure they get practical experience in a wider range of potential scenarios?

Oxford Medical Simulation provides virtual reality training to help doctors, nurses, mental health professionals, and other clinical trainees test out medical scenarios in a realistic but totally safe environment.

As trainees care for the citizens of Uncanny Valley, AI sits behind both patient behaviour and physiology, adapting in response to their clinical decisions. That means not only are trainees making medical decisions, but they can also practice their bedside manner.

3. Predictive models for understanding patient risk in nursing

A lot of healthcare discussions centre on treating illness and fixing things – the reactive component of care – but let’s talk about prevention for a minute.

Over decades of health research, clinical professionals have developed a pretty good idea of various risk factors and what they cause. Smoking leads to various cancers; high cholesterol can cause heart attacks and strokes; being overweight can put you at risk of diabetes, and so on.

But human bodies are complex, and what puts one person at risk of developing rheumatoid arthritis, for example, may lead to nothing for another. It can be difficult to predict who’s at risk of what without infinite time and budget for research – especially when there are lots of tiny interacting factors involved.

Predictive modelling can do a lot of that work for us. AI can rapidly data-mine millions of historical cases, using both numerical and natural language processing to identify common circumstances, demographics, and characteristics. The model can then be applied to any patient to assess their level of risk.

In nursing, this is already helping clinicians understand things like which factors affect the likelihood of an elderly patient suffering a fall, or what might cause an adverse reaction to medication. It has the potential to become really useful when integrated with electronic patient records, where the system can actively flag risks to a patient’s team to trigger preventative care.

4. Large language models for information sharing

Large language models (LLMs) are really having their moment – in all sorts of industries. Able to process vast amounts of text-based data and use it to inform human-like responses to prompts, LLMs are behind adaptive chatbots, services that can summarise search queries and, yes, those exceptionally famous generative AI tools.

As with all technologies, the standards for healthcare use are significantly higher. But as they mature, LLMs have the potential to extract more value from the billions of medical records, statistics, research findings, textbooks, and journal articles out there.

Applied carefully, LLMs can make it easier for researchers to access and assess existing literature, and work from larger sample sizes. Clinicians can use advanced search tools to find similar cases to help them make diagnoses or treatment decisions. And the general public can use clinically-verified chatbots to find information about their symptoms or diagnosed conditions, to help them manage their health.

5. Artificial intelligence for taking care of your own health

Speaking of managing our own health: I’d wager we’re all guilty of searching our symptoms online and going down a horrible rabbit hole of worst-case scenarios for what’s likely to be a simple headache. (I definitely went looking for images of my upcoming surgery and squeamishly regretted it.)

But there are more sensible ways of checking up on your health. Ada is a healthcare AI, designed by doctors and scientists and delivered in a handy app format, that draws on a vast, clinically-validated information base to offer assessments, triage options, and long-term tracking tools. If you’ve got symptoms that you’re worried about, but you’re not sure if you need to see a clinician, the app guides you through a questionnaire to work out what you’re likely suffering from.

With clinician shortages getting progressively worse, it may become more difficult to access services quickly, especially for minor ailments. We’d all benefit from better ways to understand our own bodies, find accurate information about our health, and decide whether that headache needs an ibuprofen or professional intervention.

Healthcare technologies are all about the people on both sides of the care equation

As B2B technology copywriters, we spend a lot of time highlighting cost savings, productivity boosts, efficiencies; business benefits that can obscure the people the technology serves.

In healthcare, it’s so much easier to draw a clear line between a technology and its value for individual people. (And, we’re all patients at one point or another.) For copywriters and B2B tech marketing audiences alike, this offers an extra connection we might not naturally feel when reading or writing about payroll software.

In healthcare, there’s no true substitute for human expertise and experience. But as AI continues to mature, there are so many clear applications where the technology can augment clinicians’ work with the right ethical and safety guidelines in place. With pervasive burnout affecting medical professionals around the world, anything that can ease their workload and support clinical decision-making is worth exploring.

Leadership agrees: in the UK, the NHS recently announced major investment in AI, with similar work going on across the EU. And the US government is investing heavily in AI for biomedical and behavioural research, with individual health systems working on their own AI deployments.

And all that means good news for me, with plenty of new technologies to learn and write about. So if you have a healthcare topic that you need to dig into, you know who to ask for.