🚀 Manish Patel

Looking for a blog recommendation? If you haven’t visited the Daysium blog yet, now’s a great time to check it out. We’ve recently explored key topics like the UK’s split year treatment and non-dom changes to support your compliance journey. I particularly recommend reading our Senior Advisor, Paul Aplin OBE's latest post. He delves into the fascinating role of digital footprints in tax compliance—definitely worth a read. Here’s the link to Paul’s article: https://lnkd.in/e8Ns_8ZT I’d love to hear about your favourite blogs too, so feel free to share!

6

As the Chief Technology Officer at a Health startup using AI, I see the introduction of the "Artificial Intelligence (Regulation) Bill" as a critical juncture for the technology sector. This proposed legislation offers a chance to establish a regulatory framework that not only protects the public and maintains ethical standards but also encourages innovation. The creation of an AI Authority could serve as a centralised body to ensure consistency and coherence across various sectors, effectively addressing the current regulatory fragmentation. Such regulation is crucial to balance rapid technological advancements with the need for transparency and accountability, ultimately fostering a more trustworthy AI ecosystem. This bill could set a precedent for global AI governance, positioning the UK as a leader in this vital area. Lord Holmes #aibill #ArtificialIntelligence #startups #ai #ukai #aijobs

13

1 Comment

This is another useful proof point of the need for #AI re-skilling across all companies and organisations. Siobhan's list of her own re-skilling and educational activities is a tribute to her inquisitive mind and resilience. It's clear that there's some great material out there from the big tech players and leading universities. What is less clear is what you need to do - at a personal or organisational level - to re-skill and reposition yourself. There's simply so much out there. With this in mind, my co-founder Shivika Pratap and I are completely reshaping https://HeyDeesha.ai to focus on guiding careers in the context of the AI economy. We are doing our reshaping based on a taxonomy of roles that we see emerging. Please DM me if you are interested in what the taxonomy looks like. It's not just tech roles, BTW. We want to extend the taxonomy to Marketing, Sales, Support... and, one day, really deep specialist skills such as Medicine. We are also looking at how the pace of change in the workplace is affecting maternity leavers/returners. Kate Apaflo Awuku-Darko, Emmanuel Ihenacho

10

3 Comments

You may not have seen Alex Hern's article about AI's large language models (LLMs) #reversalcurse. The article nicely framed the example of Tom Cruise's mother (and the lesser-known 'token'—Mary Lee Pfeiffer’s son) and the difficulty LLMs have with reasoning (e.g reverse logic). I have a confession: I read the three puzzles Alex suggested and also made a mistake with the game-show-three-door-pick-problem! Grr. I assumed I knew the riddle without paying attention. In my case, I returned the incorrect answer through carelessness rather than my inability to reason. (I guess humans have their downsides too - although I speak for myself rather than for the rest of humanity) This quote from (the positively sceptical) Gary Marcus sums it up best: "The median split of AI wisdom is this: either you understand that current neural networks struggle mightily with outliers (just as their 1990s predecessors did) – and therefore understand why current AI is doomed to fail on many of its most lavish promises – or you don’t." My only question to Gary and Alex - is it really the 'median' split of AI wisdom?

1

There's been some great talks on advancements in big data and AI at the Dementias Platform UK Translation 2024 conference today. Key takeaways: 1) Standardisation in big data is key to developing AI models in health - DPUK has worked on developing pipelines for the standardisation, anonymisation, and processing of complex big data in health to provide research ready data. See our preprint here - https://lnkd.in/ex9GSPfM 2) We need more representative datasets to avoid bias and discrimination in AI models - DPUK is working on building international collaborations and building our cohort collections to include a diverse range of populations for training and validating AI. 3) The risks of AI models need to be carefully considered and mitigated - DPUK has been assessing privacy concerns in AI models and how we can help researchers mitigate these. Report to be published next week! 4) Synthetic data offers ways to improve robustness in AI models for rare diseases - my PhD project is looking at developing generative models for creating synthetic brain imaging data of less common dementia types.

11

This morning on BBC News the UK Health Secretary, Wes Streeting, highlighted the urgent need for digital transformation in the NHS . In my latest blog post, I explore how AI-powered digital pathology can help the NHS achieve its goals of faster diagnoses and a shift towards preventative care. https://lnkd.in/enqbCH8h   In this blog, I discuss how AI can streamline pathology workflows, leading to more efficient and accurate diagnoses. It's NOT about replacing pathologists but empowering them to focus on what matters most - the patients.   As I mentioned in Pathology in Practice a couple of days ago, this technology is already making a difference. But we need continued collaboration to scale its impact. #AI #healthcare #digitaltransformation #NHS #pathology #patientcare

25

One of the best blogs I've come across on quantization, with outstanding visualizations. Highly recommend a read for anyone looking to understand LLM quantization in more depth! 📐 - https://lnkd.in/e_WTgq6x At Felafax AI (YC S24), we are building Infra to make full precision multi-GPU fine-tuning easier. #AI #Infra #LLM

16

3 Comments

This TechCrunch article highlights 13.5 million hours in the NHS lost due to poor IT systems. I can certainly echo that issue with my recent experiences that show a fractured and ineffective use of IT and poor IT policies. (I won't bore you with the details, but ask me directly if you want to know). I believe that better IT could be a massive opportunity for the NHS. Yes, it would mean spending to save, but there's a significant opportunity to empower individuals, while eliminating much of the NHS bureaucracy and overheads. The greatest challenge ahead is two-fold. Firstly finding someone with the vision to transform NHS IT systems. Second, the ability to execute that vision without giving away massive, poorly structured and expensive contracts to the usual suspects. Would it be unreasonable for the NHS to create an internal start-up to develop new technology? Developers in Silicon Valley can create new companies with only a few millions in investment. So, should the government fund multiple skunkworks projects and see what falls out? Who will be the NHS IT Czar capable of completing a vision of the future? https://lnkd.in/erhX6cKR

6

1 Comment

https://lnkd.in/eWK3ZVsM RAGs have become the standard architecture for providing LLMs with context in order to avoid hallucinations. However even RAGs can suffer from hallucination, as is often the case when the retrieval fails to retrieve sufficient context or even retrieves irrelevant context that is then weaved into the LLM’s response.

4

🧪Revolutionising Cancer Care: AI Predicts Personalised Treatment Outcomes🔬 Some good news for Friday! 🚀 Exciting strides in AI at the National Institutes of Health (NIH). Researchers are now harnessing AI to predict how patients will respond to specialised cancer treatments, notably immunotherapy, which currently benefits only 20% of patients. 💉 Using simple blood tests, the AI model accounts for factors like age and cancer type to determine the best treatment options without invasive procedures. Even more impressive, it can predict patient survival outcomes in both the short and long term. 🔍 This breakthrough exemplifies AI's transformative potential in healthcare, joining innovations like AI-powered stethoscopes and intelligent medical scans. With the AI healthcare market poised to hit $188 billion by 2030, the future looks bright.

2

I have long been interested in the potential of this technology

6

2 Comments

71% of patients lack confidence that health records are stored securely in the UK Blockchain can transform how patients control their data, how healthcare providers verify qualifications, ensure regulatory compliance, and authenticate pharmaceutical products—ultimately contributing to a more reliable and efficient healthcare system. At the forefront of this transformation is decentralized, verifiable data powered by blockchain technology based on Self-Sovereign Identity (SSI) principles. Visit our stand and experience how enhancing data security, ensuring patient privacy, and streamlining processes can address the pressing challenges in healthcare. Empeiria's infrastructure can support a more transparent and effective healthcare ecosystem, paving the way for improved health outcomes and heightened patient satisfaction. #blockchain #healthcare #ksa #GHE24 #GlobalHealthExpo

25

11 Comments

🚀 **Unleashing the Power of Federated Data in Computational Biology** 🚀 Leaders in Biopharma and Biotech, it's time for a paradigm shift in how we approach data management. Fragmented data, duplicate data vendors, and persistent bottlenecks are stifling innovation. The solution? A comprehensive Federated Data Set. Here’s why: 🔍 **Tackle Fragmented Data:** - Seamless integration across diverse datasets. - Enables holistic insights by bridging the silos. - Achieves higher data consistency and reliability. 💡 **Eliminate Duplicate Vendors:** - Streamlines data sourcing processes. - Reduces redundant contracts and costs. - Simplifies vendor management while maintaining high data quality. ⏱️ **Resolve Data Bottlenecks:** - Speeds up data retrieval and analysis. - Facilitates real-time data access and sharing. - Enhances collaboration across research teams and departments. 📈 **Benefits of Federated Data in Computational Biology:** - **Holistic Insights:** Combines data from genomics, proteomics, and clinical trials for a holistic view. - **Accelerated Innovation:** Faster analysis leads to rapid hypothesis testing and drug discovery. - **Enhanced Collaboration:** Promotes cross-functional teamwork, driving efficiency and groundbreaking results. - **Regulatory Compliance:** Ensures data governance standards are met effortlessly. 🚨 **Call to Action:** - Embrace federated data strategies to stay ahead in the competitive landscape. - Partner with industry leaders to implement best practices. - Invest in technologies that streamline data integration and accessibility. Let's transform how we harness the power of data in Computational Biology. The future of biopharma and biotech innovation relies on our ability to manage and leverage data efficiently. #biopharma #biotech #computationalbiology #Innovation

3

The simplest way to get your study documents organised is with an eTMF vault. There is just so much to remember to stay on top of your game with, potentially, hundreds of docs per trial - PDFs, spreadsheets, contracts, agreements, SOPs... the list goes on. Typical things I hear in this space are: 👉🏼 What documents do I need and when? 👉🏼 Who needs to be aware, access and sign certain docs? 👉🏼 How do I keep on top of version numbers? 👉🏼 I'm new to this - I can't keep up with all the acronyms! 👉🏼 What if I miss something? If you're adhering to best practices like ICH-GCP you'll want to do the right thing from the outset to avoid many headaches. Learn more here: https://buff.ly/4cyX321 #clinicaltrials #nutritionaltrials #humantrials #epro #ecrf #econsent #etmf #cognitivetests #wearables #randomization #ctms #decentralisedclinicaltrials #trialflare

2

A great case study, detailing how The Duckhorn Portfolio are using Claret, resulting in significant increases in forecast accuracy. Check it out (perhaps whilst enjoying a glass of one of Duckhorn's wines...) https://lnkd.in/e-kVPR9C #sop #ibp #demandplanning

4

2 Comments

Fantastic day at Big Data LDN 2024! I had the chance to connect with some brilliant minds in the industry. Special thanks to my good friend Andrew Carruthers for translating deep technical content & Chris Tabb for hosting the low-key insightful meetup—plenty of great (and spicy) discussions around the intersection of data and AI. It’s exciting to see how far the field has come, and the conversations confirmed that we’re just getting started. Big things to follow with Linnk Group.... #BigDataLDN #AI #DataStrategy #Networking #linnkgroup #linnk

49

3 Comments

Following on from my post on vectorising Wikipedia yesterday, here's a blog containing links to the files. Have fun with it, and let me know if you find any interesting uses for the data!

8

1 Comment