Healthcare saw significant disruptions four years ago, and while we will continue to feel the impact for the next decade, one positive outcome was witnessing the industry’s ability to transform amid such immense pressure. I am definitely not suggesting another healthcare calamity! But disruptions can have a positive impact – and I believe that technology will continue to disrupt healthcare at pace. Recently, my colleague Tim Sheedy shared his thoughts on how 2024 is poised to become the year of the AI startup, highlighting innovative options that organisations should consider in their AI journeys. AI startups and innovators hold the potential to further the “good disruption” that will transform healthcare.

Of course, there are challenges associated, including concerns on ethical and privacy-related issues, the reliability of technology – particularly while scaling – and on professional liability. However, the industry cannot overlook the substantial number of innovative startups that are using AI technologies to address some of the most pressing challenges in the healthcare industry.

Why Now?

AI is not new to healthcare. Many would cite the development of MYCIN – an early AI program aimed at identifying treatments for blood infections – as the first known example. It did kindle interest in research in AI and even during the 1980s and 1990s, AI brought about early healthcare breakthroughs, including faster data collection and processing, enhanced precision in surgical procedures, and research and mapping of diseases.

Now, healthcare is at an AI inflection point due to a convergence of three significant factors.

• Advanced AI. AI algorithms and capabilities have become more sophisticated, enabling them to handle complex healthcare data and tasks with greater accuracy and efficiency.
• Demand for Accessible Healthcare. Healthcare systems globally are striving for better care amid resource constraints, turning to AI for efficiency, cost reduction, and broader access.
• Consumer Demand. As people seek greater control over their health, personalised care has become essential. AI can analyse vast patient data to identify health risks and customise care plans, promoting preventative healthcare.

Promising Health AI Startups

As innovative startups continue to emerge in healthcare, we’re particularly keeping an eye on those poised to revolutionise diagnostics, care delivery, and wellness management. Here are some examples.

DIAGNOSTICS

• Claritas HealthTech has created advanced image enhancement software to address challenges in interpreting unclear medical images, improving image clarity and precision. A cloud-based platform with AI diagnostic tools uses their image enhancement technology to achieve greater predictive accuracy.
• Ibex offers Galen, a clinical-grade, multi-tissue platform to detect and grade cancers, that integrate with third-party digital pathology software solutions, scanning platforms, and laboratory information systems.
• MEDICAL IP is focused on advancing medical imaging analysis through AI and 3D technologies (such as 3D printing, CAD/CAM, AR/VR) to streamline medical processes, minimising time and costs while enhancing patient comfort.
• Verge Genomics is a biopharmaceutical startup employing systems biology to expedite the development of life-saving treatments for neurodegenerative diseases. By leveraging patient genomes, gene expression, and epigenomics, the platform identifies new therapeutic gene targets, forecasts effective medications, and categorises patient groups for enhanced clinical efficacy.
• X-Zell focuses on advanced cytology, diagnosing diseases through single atypical cells or clusters. Their plug-and-play solution detects, visualises, and digitises these phenomena in minimally invasive body fluids. With no complex specimen preparation required, it slashes the average sample-to-diagnosis time from 48 hours to under 4 hours.

CARE DELIVERY

• Abridge specialises in automating clinical notes and medical discussions for physicians, converting patient-clinician conversations into structured clinical notes in real time, powered by GenAI. It integrates seamlessly with EMRs such as Epic.
• Waltz Health offers AI-driven marketplaces aimed at reducing costs and innovative consumer tools to facilitate informed care decisions. Tailored for payers, pharmacies, and consumers, they introduce a fresh approach to pricing and reimbursing prescriptions that allows consumers to purchase medication at the most competitive rates, improving accessibility.
• Acorai offers a non-invasive intracardiac pressure monitoring device for heart failure management, aimed at reducing hospitalisations and readmissions. The technology can analyse acoustics, vibratory, and waveform data using ML to monitor intracardiac pressures.

WELLNESS MANAGEMENT

• Anya offers AI-driven support for women navigating life stages such as fertility, pregnancy, parenthood, and menopause. For eg. it provides support during the critical first 1,001 days of the parental journey, with personalised advice, tracking of developmental milestones, and connections with healthcare professionals.
• Dacadoo’s digital health engagement platform aims to motivate users to adopt healthier lifestyles through gamification, social connectivity, and personalised feedback. By analysing user health data, AI algorithms provide tailored insights, goal-setting suggestions, and challenges.

Conclusion

There is no question that innovative startups can solve many challenges for the healthcare industry. But startups flourish because of a supportive ecosystem. The health innovation ecosystem needs to be a dynamic network of stakeholders committed to transforming the industry and health outcomes – and this includes healthcare providers, researchers, tech companies, startups, policymakers, and patients. Together we can achieve the longstanding promise of accessible, cost-effective, and patient-centric healthcare.

However, legacy modernisation and migration initiatives carry significant risks.  For instance, TSB faced a USD 62M fine due to a failed mainframe migration, resulting in severe disruptions to branch operations and core banking functions like telephone, online, and mobile banking. The migration failure led to 225,492 complaints between 2018 and 2019, affecting all 550 branches and required TSB to pay more than USD 25M to customers through a redress program.

Modernisation Options

• Rip and Replace. Replacing the entire legacy system with a modern, cloud-based solution. While offering a clean slate and faster time to value, it’s expensive, disruptive, and carries migration risks.
• Refactoring. Rewriting key components of the legacy system with modern languages and architectures. It’s less disruptive than rip-and-replace but requires skilled developers and can still be time-consuming.
• Encapsulation. Wrapping the legacy system with a modern API layer, allowing integration with newer applications and tools. It’s quicker and cheaper than other options but doesn’t fully address underlying limitations.
• Microservices-based Modernisation. Breaking down the legacy system into smaller, independent services that can be individually modernised over time. It offers flexibility and agility but requires careful planning and execution.

Financial Systems on the Block for Legacy Modernisation

Data Analytics Platforms. Harnessing customer data for insights and targeted offerings is vital. Legacy data warehouses often struggle with real-time data processing and advanced analytics.

CRM Systems. Effective customer interactions require integrated CRM platforms. Outdated systems might hinder communication, personalisation, and cross-selling opportunities.

Payment Processing Systems. Legacy systems might lack support for real-time secure transactions, mobile payments, and cross-border transactions.

Core Banking Systems (CBS). The central nervous system of any bank, handling account management, transactions, and loan processing. Many Asia Pacific banks rely on aging, monolithic CBS with limited digital capabilities.

Digital Banking Platforms. While several Asia Pacific banks provide basic online banking, genuine digital transformation requires mobile-first apps with features such as instant payments, personalised financial management tools, and seamless third-party service integration.

Modernising Technical Approaches and Architectures

Numerous technical factors need to be addressed during modernisation, with decisions needing to be made upfront. Questions around data migration, testing and QA, change management, data security and development methodology (agile, waterfall or hybrid) need consideration.

Best practices in legacy migration have taught some lessons.

Adopt a data fabric platform. Many organisations find that centralising all data into a single warehouse or platform rarely justifies the time and effort invested. Businesses continually generate new data, adding sources, and updating systems. Managing data where it resides might seem complex initially. However, in the mid to longer term, this approach offers clearer benefits as it reduces the likelihood of data discrepancies, obsolescence, and governance challenges.

Focus modernisation on the customer metrics and journeys that matter. Legacy modernisation need not be an all-or-nothing initiative. While systems like mainframes may require complete replacement, even some mainframe-based software can be partially modernised to enable services for external applications and processes. Assess the potential of modernising components of existing systems rather than opting for a complete overhaul of legacy applications.

Embrace the cloud and SaaS. With the growing network of hyperscaler cloud locations and data centres, there’s likely to be a solution that enables organisations to operate in the cloud while meeting data residency requirements. Even if not available now, it could align with the timeline of a multi-year legacy modernisation project. Whenever feasible, prioritise SaaS over cloud-hosted applications to streamline management, reduce overhead, and mitigate risk.

Build for customisation for local and regional needs. Many legacy applications are highly customised, leading to inflexibility, high management costs, and complexity in integration. Today, software providers advocate minimising configuration and customisation, opting for “out-of-the-box” solutions with room for localisation. The operations in different countries may require reconfiguration due to varying regulations and competitive pressures. Architecting applications to isolate these configurations simplifies system management, facilitating continuous improvement as new services are introduced by platform providers or ISV partners.

Explore the opportunity for emerging technologies. Emerging technologies, notably AI, can significantly enhance the speed and value of new systems. In the near future, AI will automate much of the work in data migration and systems integration, reducing the need for human involvement. When humans are required, low-code or no-code tools can expedite development. Private 5G services may eliminate the need for new network builds in branches or offices. AIOps and Observability can improve system uptime at lower costs. Considering these capabilities in platform decisions and understanding the ecosystem of partners and providers can accelerate modernisation journeys and deliver value faster.

Don’t Let Analysis Paralysis Slow Down Your Journey!

Yes, there are a lot of decisions that need to be made; and yes, there is much at stake if things go wrong! However, there’s a greater risk in not taking action. Maintaining a laser-focus on the customer and business outcomes that need to be achieved will help align many decisions. Keeping the customer experience as the guiding light ensures organisations are always moving in the right direction.

In this eBook, Ecosystm VP Research Tim Sheedy and Vinod Bijlani and Aman Deep from HPE APAC share their insights on why it is crucial to establish tailored AI capabilities within the organisation.

AI-Powered Enterprise_HPE_Ecosystm_eBook-1

AI-Powered Enterprise_HPE_Ecosystm_eBook_2

AI-Powered Enterprise_HPE_Ecosystm_eBook-3

AI-Powered Enterprise_HPE_Ecosystm_eBook-4

AI-Powered Enterprise_HPE_Ecosystm_eBook-5

AI-Powered Enterprise_HPE_Ecosystm_eBook-6

AI-Powered Enterprise_HPE_Ecosystm_eBook-7

AI-Powered Enterprise_HPE_Ecosystm_eBook-8

AI-Powered Enterprise_HPE_Ecosystm_eBook-9

AI-Powered Enterprise_HPE_Ecosystm_eBook-10

AI-Powered Enterprise_HPE_Ecosystm_eBook-11

AI-Powered Enterprise_HPE_Ecosystm_eBook-12

Click here to download the eBook “AI-Powered Enterprise: Building a Data Driven Foundation To Super Charge Your AI Journey”

The International Energy Agency (IEA) estimates that data centres are responsible for up to 1.5% of global electricity use and 1% of GHG emissions. Cloud providers have recognised this and are committed to change. Between 2025 and 2030, all hyperscalers – AWS, Azure, Google, and Oracle included – expect to power their global cloud operations entirely with renewable sources.

Chasing the Sun

Cloud providers are shifting their sights from simply matching electricity use with renewable power purchase agreements (PPA) to the more ambitious goal of operating 24/7 on carbon-free sources. A defining characteristic of renewables though is intermittency, with production levels fluctuating based on the availability of sunlight and wind. Leading cloud providers are using AI to dynamically distribute compute workloads throughout the day to regions with lower carbon intensity. Workloads that are processed with solar power during daylight can be shifted to nearby regions with abundant wind energy at night.

Addressing Water Scarcity

Many of the largest cloud data centres are situated in sunny locations to take advantage of solar power and proximity to population centres. Unfortunately, this often means that they are also in areas where water is scarce. While liquid-cooled facilities are energy efficient, local communities are concerned on the strain on water sources. Data centre operators are now committing to reduce consumption and restore water supplies. Simple measures, such as expanding humidity (below 20% RH) and temperature tolerances (above 30°C) in server rooms have helped companies like Meta to cut wastage. Similarly, Google has increased their reliance on non-potable sources, such as grey water and sea water.

From Waste to Worth

Data centre operators have identified innovative ways to reuse the excess heat generated by their computing equipment. Some have used it to heat adjacent swimming pools while others have warmed rooms that house vertical farms. Although these initiatives currently have little impact on the environmental impact of cloud, they suggest a future where waste is significantly reduced.

Greening the Grid

The giant facilities that cloud providers use to house their computing infrastructure are also set to change. Building materials and construction account for an astonishing 11% of global carbon emissions. The use of recycled materials in concrete and investing in greener methods of manufacturing steel are approaches the construction industry are attempting to lessen their impact. Smaller data centres have been 3D printed to accelerate construction and use recyclable printing concrete. While this approach may not be suitable for hyperscale facilities, it holds potential for smaller edge locations.

Rethinking Hardware Management

Cloud providers rely on their scale to provide fast, resilient, and cost-effective computing. In many cases, simply replacing malfunctioning or obsolete equipment would achieve these goals better than performing maintenance. However, the relentless growth of e-waste is putting pressure on cloud providers to participate in the circular economy. Microsoft, for example, has launched three Circular Centres to repurpose cloud equipment. During the pilot of their Amsterdam centre, it achieved 83% reuse and 17% recycling of critical parts. The lifecycle of equipment in the cloud is largely hidden but environmentally conscious users will start demanding greater transparency.

Recommendations

Organisations should be aware of their cloud-derived scope 3 emissions and consider broader environmental issues around water use and recycling. Here are the steps that can be taken immediately:

1. Monitor GreenOps. Cloud providers are adding GreenOps tools, such as the AWS Customer Carbon Footprint Tool, to help organisations measure the environmental impact of their cloud operations. Understanding the relationship between cloud use and emissions is the first step towards sustainable cloud operations.
2. Adopt Cloud FinOps for Quick ROI. Eliminating wasted cloud resources not only cuts costs but also reduces electricity-related emissions. Tools such as CloudVerse provide visibility into cloud spend, identifies unused instances, and helps to optimise cloud operations.
3. Take a Holistic View. Cloud providers are being forced to improve transparency and reduce their environmental impact by their biggest customers. Getting educated on the actions that cloud partners are taking to minimise emissions, water use, and waste to landfill is crucial. In most cases, dedicated cloud providers should reduce waste rather than offset it.
4. Enable Remote Workforce. Cloud-enabled security and networking solutions, such as SASE, allow employees to work securely from remote locations and reduce their transportation emissions. With a SASE deployed in the cloud, routine management tasks can be performed by IT remotely rather than at the branch, further reducing transportation emissions.

In the Ecosystm Predicts: Building an Agile & Resilient Organisation: Top 5 Trends in 2024​, Principal Advisor Darian Bird said, “The emergence of Generative AI combined with the maturing of deepfake technology will make it possible for malicious agents to create personalised voice and video attacks.” Darian highlighted that this democratisation of phishing, facilitated by professional-sounding prose in various languages and tones, poses a significant threat to potential victims who rely on misspellings or oddly worded appeals to detect fraud. As we see more of these attacks and social engineering attempts, it is important to improve defence mechanisms and increase awareness. 

Understanding Deepfake Technology 

The term Deepfake is a combination of the words ‘deep learning’ and ‘fake’. Deepfakes are AI-generated media, typically in the form of images, videos, or audio recordings. These synthetic content pieces are designed to appear genuine, often leading to the manipulation of faces and voices in a highly realistic manner. Deepfake technology has gained spotlight due to its potential for creating convincing yet fraudulent content that blurs the line of reality. 

Deepfake algorithms are powered by Generative Adversarial Networks (GANs) and continuously enhance synthetic content to closely resemble real data. Through iterative training on extensive datasets, these algorithms refine features such as facial expressions and voice inflections, ensuring a seamless emulation of authentic characteristics.  

Deepfakes Becoming Increasingly Convincing 

Hyper-realistic deepfakes, undetectable to the human eye and ear, have become a huge threat to the financial and technology sectors. Deepfake technology has become highly convincing, blurring the line between real and fake content. One of the early examples of a successful deepfake fraud was when a UK-based energy company lost USD 243k through a deepfake audio scam in 2019, where scammers mimicked the voice of their CEO to authorise an illegal fund transfer.  

Deepfakes have evolved from audio simulations to highly convincing video manipulations where faces and expressions are altered in real-time, making it hard to distinguish between real and fake content. In 2022, for instance, a deepfake video of Elon Musk was used in a crypto scam that resulted in a loss of about USD 2 million for US consumers. This year, a multinational company in Hong Kong lost over USD 25 million when an employee was tricked into sending money to fraudulent accounts after a deepfake video call by what appeared to be his colleagues. 

Regulatory Responses to Deepfakes 

Countries worldwide are responding to the challenges posed by deepfake technology through regulations and awareness campaigns. 

• Singapore’s Online Criminal Harms Act, that will come into effect in 2024, will empower authorities to order individuals and Internet service providers to remove or block criminal content, including deepfakes used for malicious purposes.  
• The UAE National Programme for Artificial Intelligence released a deepfake guide to educate the public about both harmful and beneficial applications of this technology. The guide categorises fake content into shallow and deep fakes, providing methods to detect deepfakes using AI-based tools, with a focus on promoting positive uses of advanced technologies. 
• The proposed EU AI Act aims to regulate them by imposing transparency requirements on creators, mandating them to disclose when content has been artificially generated or manipulated. 
• South Korea passed a law in 2020 banning the distribution of harmful deepfakes. Offenders could be sentenced to up to five years in prison or fined up to USD 43k. 
• In the US, states like California and Virginia have passed laws against deepfake pornography, while federal bills like the DEEP FAKES Accountability Act aim to mandate disclosure and counter malicious use, highlighting the diverse global efforts to address the multifaceted challenges of deepfake regulation. 

Detecting and Protecting Against Deepfakes 

Detecting deepfake becomes increasingly challenging as technology advances. Several methods are needed – sometimes in conjunction – to be able to detect a convincing deepfake. These include visual inspection that focuses on anomalies, metadata analysis to examine clues about authenticity, forensic analysis for pattern and audio examination, and machine learning that uses algorithms trained on real and fake video datasets to classify new videos.  

However, identifying deepfakes requires sophisticated technology that many organisations may not have access to. This heightens the need for robust cybersecurity measures. Deepfakes have seen an increase in convincing and successful phishing – and spear phishing – attacks and cyber leaders need to double down on cyber practices.  

Defences can no longer depend on spotting these attacks. It requires a multi-pronged approach which combines cyber technologies, incidence response, and user education.  

Preventing access to users. By employing anti-spoofing measures organisations can safeguard their email addresses from exploitation by fraudulent actors. Simultaneously, minimising access to readily available information, particularly on websites and social media, reduces the chance of spear-phishing attempts. This includes educating employees about the implications of sharing personal information and clear digital footprint policies. Implementing email filtering mechanisms, whether at the server or device level, helps intercept suspicious emails; and the filtering rules need to be constantly evaluated using techniques such as IP filtering and attachment analysis.  

Employee awareness and reporting. There are many ways that organisations can increase awareness in employees starting from regular training sessions to attack simulations. The usefulness of these sessions is often questioned as sometimes they are merely aimed at ticking off a compliance box. Security leaders should aim to make it easier for employees to recognise these attacks by familiarising them with standard processes and implementing verification measures for important email requests. This should be strengthened by a culture of reporting without any individual blame. 

Securing against malware. Malware is often distributed through these attacks, making it crucial to ensure devices are well-configured and equipped with effective endpoint defences to prevent malware installation, even if users inadvertently click on suspicious links. Specific defences may include disabling macros and limiting administrator privileges to prevent accidental malware installation. Strengthening authentication and authorisation processes is also important, with measures such as multi-factor authentication, password managers, and alternative authentication methods like biometrics or smart cards. Zero trust and least privilege policies help protect organisation data and assets.   

Detection and Response. A robust security logging system is crucial, either through off-the shelf monitoring tools, managed services, or dedicated teams for monitoring. What is more important is that the monitoring capabilities are regularly updated. Additionally, having a well-defined incident response can swiftly mitigate post-incident harm post-incident. This requires clear procedures for various incident types and designated personnel for executing them, such as initiating password resets or removing malware. Organisations should ensure that users are informed about reporting procedures, considering potential communication challenges in the event of device compromise. 

Conclusion 

The rise of deepfakes has brought forward the need for a collaborative approach. Policymakers, technology companies, and the public must work together to address the challenges posed by deepfakes. This collaboration is crucial for making better detection technologies, establishing stronger laws, and raising awareness on media literacy. 

In my last Ecosystm Insights, I spoke about the implications of the shift from Predictive AI to Generative AI on ROI considerations of AI deployments. However, from my discussions with colleagues and technology leaders it became clear that there is a need to define and distinguish between Predictive AI and Generative AI better.

Predictive AI analyses historical data to predict future outcomes, crucial for informed decision-making and strategic planning. Generative AI unlocks new avenues for innovation by creating novel data and content. Organisations need both – Predictive AI for enhancing operational efficiencies and forecasting capabilities and Generative AI to drive innovation; create new products, services, and experiences; and solve complex problems in unprecedented ways. 

This guide aims to demystify these categories, providing clarity on their differences, applications, and examples of the algorithms they use. 

Predictive AI: Forecasting the Future

Predictive AI is extensively used in fields such as finance, marketing, healthcare and more. The core idea is to identify patterns or trends in data that can inform future decisions. Predictive AI relies on statistical, machine learning, and deep learning models to forecast outcomes. 

Key Algorithms in Predictive AI 

• Regression Analysis. Linear and logistic regression are foundational tools for predicting a continuous or categorical outcome based on one or more predictor variables. 
• Decision Trees. These models use a tree-like graph of decisions and their possible consequences, including chance event outcomes, resource costs and utility. 
• Random Forest (RF). An ensemble learning method that operates by constructing a multitude of decision trees at training time to improve predictive accuracy and control over-fitting. 
• Gradient Boosting Machines (GBM). Another ensemble technique that builds models sequentially, each new model correcting errors made by the previous ones, used for both regression and classification tasks. 
• Support Vector Machines (SVM). A supervised machine learning model that uses classification algorithms for two-group classification problems. 

Generative AI: Creating New Data

Generative AI, on the other hand, focuses on generating new data that is similar but not identical to the data it has been trained on. This can include anything from images, text, and videos to synthetic data for training other AI models. GenAI is particularly known for its ability to innovate, create, and simulate in ways that predictive AI cannot. 

Key Algorithms in Generative AI 

• Generative Adversarial Networks (GANs). Comprising two networks – a generator and a discriminator – GANs are trained to generate new data with the same statistics as the training set. 
• Variational Autoencoders (VAEs). These are generative algorithms that use neural networks for encoding inputs into a latent space representation, then reconstructing the input data based on this representation. 
• Transformer Models. Originally designed for natural language processing (NLP) tasks, transformers can be adapted for generative purposes, as seen in models like GPT (Generative Pre-trained Transformer), which can generate coherent and contextually relevant text based on a given prompt. 

Comparing Predictive and Generative AI

The fundamental difference between the two lies in their primary objectives: Predictive AI aims to forecast future outcomes based on past data, while Generative AI aims to create new, original data that mimics the input data in some form. 

The differences become clearer when we look at these examples.  

Predictive AI Examples  

• Supply Chain Management. Analyses historical supply chain data to forecast demand, manage inventory levels, reduces costs and improve delivery times.  
• Healthcare. Analysing patient records to predict disease outbreaks or the likelihood of a disease in individual patients. 
• Predictive Maintenance. Analyse historical and real-time data and preemptively identifies system failures or network issues, enhancing infrastructure reliability and operational efficiency. 
• Finance. Using historical stock prices and indicators to predict future market trends. 

Generative AI Examples  

• Content Creation. Generating realistic images or art from textual descriptions using GANs. 
• Text Generation. Creating coherent and contextually relevant articles, stories, or conversational responses using transformer models like GPT-3. 
• Chatbots and Virtual Assistants. Advanced GenAI models are enhancing chatbots and virtual assistants, making them more realistic. 
• Automated Code Generation. By the use of natural language descriptions to generate programming code and scripts, to significantly speed up software development processes. 

Conclusion 

Organisations that exclusively focus on Generative AI may find themselves at the forefront of innovation, by leveraging its ability to create new content, simulate scenarios, and generate original data. However, solely relying on Generative AI without integrating Predictive AI’s capabilities may limit an organisation’s ability to make data-driven decisions and forecasts based on historical data. This could lead to missed opportunities to optimise operations, mitigate risks, and accurately plan for future trends and demands. Predictive AI’s strength lies in analysing past and present data to inform strategic decision-making, crucial for long-term sustainability and operational efficiency. 

It is essential for companies to adopt a dual-strategy approach in their AI efforts. Together, these AI paradigms can significantly amplify an organisation’s ability to adapt, innovate, and compete in rapidly changing markets.