Digital Assessment Methods

🔥 Framework for selecting digital health maturity assessments 📢 The Global Digital Health Partnership (GDHP) has published a paper, covering a comprehensive review of existing digital health maturity assessments, analysing the levels at which they are applied and the specific areas they evaluate. This has been further strengthened by insights from an international survey of GDHP member countries, highlighting not only what is currently valued in maturity models but also what is needed next.   🔔 The publication identifies five core meta-domains that underpin digital health maturity: 🔹 Leadership, strategy, policy, and guidance 🔹 Infrastructure, operations, and financial management 🔹 Data and analytics for health 🔹 User attitudes, capability, adoption, and feedback 🔹 Quality improvement and outcome tracking   Each meta-domain can be further broken down into sub-domains, such as data governance, cybersecurity preparedness, health workforce capability or user attitudes towards digital health. 📊 The paper includes an overview of 44 digital health maturity assessment models, including their assessment level, health sector coverage and evaluation ratings according to the proposed evaluation framework. 🪞 GDHP member countries are more focused on their own digital health strategies and outcomes, than evaluating how their health systems compare Digital health maturity assessments support them in developing digital health legislation and policies, in identifying priority areas to address and in making appropriate investments in digital health initiatives. At the same time, there is the feeling that existing assessments can't be effectively applied to the current digital health landscape. 👣 The GDHP feedback indicates that there is a need for an evaluation framework for digital health maturity assessments. The importance of local relevance and contextual knowledge also emerged as a top priority for many countries. Therefore, next steps to further advance this work could for instance include an analysis of domain coverage in the identified digital health maturity models by assessment level – at national, subnational and organizational level - and health sector coverage, or a gap analysis to identify which global regions may be missing or which meta-domains or sub-domains may need more focus. #digitalhealth #maturity #framework

For years, we’ve struggled to truly understand what’s happening in the brain, especially in the earliest stages of cognitive decline.   Despite decades of investment, Alzheimer’s clinical trials more often than not have fallen short. Not because the science isn’t promising, but because the tools investigators relied on weren’t built for early detection or real-world insight. They weren’t built for today.   Assessments like MMSE, ADAS-cog and CDR sum of boxes were designed to confirm decline, not to catch it early or track meaningful change over time. These ill-suited tools often miss the subtle, functional shifts that matter most. And now, as new treatments emerge, they’re often reaching patients too late to make a difference.   This problem isn't theoretical. Recent promising programs from Cassava Sciences , Annovis Bio, Inc., and UCB have missed their primary endpoints—despite strong scientific rationales. Why? Possibly because they relied on traditional measures that are too blunt, too slow, and too subjective to detect meaningful change early enough.   How can 21st-century therapies succeed if we're still measuring them with inadequate 20th-century endpoints?   With millions of people affected and new therapies finally reaching the market, every missed signal is a missed opportunity. After decades of failed trials, the industry is demanding smarter ways to prove efficacy and bring treatments to patients faster.   The good news: we're entering an era where cognitive and functional changes can be captured through digital, multimodal data. Advances in #AI, #AR, and #sensorfusion now enable assessments that reflect how people function in the real world.   This shift is already underway, and Altoida, Inc. has been fortunate to help define it.   The Altoida Digital NeuroMarker Platform is already supporting pharma and research teams like Eisai Co., Ltd.  Johnson & Johnson Innovative Medicine who are rethinking how cognitive change is measured—moving from low-fidelity snapshots to continuous, high-resolution insights. What we’re seeing is a clearer, earlier, and more actionable view of brain health, one that could significantly improve the success of therapies already in development.   The results speak for themselves: over 40,000 assessments completed across nearly 20,000 participants. And independent, peer-reviewed studies consistently show our approach detects cognitive decline earlier than traditional tools, with greater sensitivity and real-world relevance.   But clinical trials are only the beginning.   This is how the next generation of #Alzheimer’s care begins— with better tools, deployed earlier, closer to the patient.   Altoida’s Digital NeuroMarker Platform is being built to unlock a new axis of intervention: enabling diagnosis in frontline care, informing treatment decisions and monitoring outcomes over time.     We’re ushering in a new era of Alzheimer’s care, from trial optimization to front-line diagnosis.    Watch what’s next.

Students need better ways to express what they’ve learned - and who they are. While traditional assessments like quizzes and essays still have value, these formats don’t always let students express their creativity or individuality. The Subject team is rethinking how students demonstrate their understanding by introducing video and audio submissions - formats that encourage creativity and give students the opportunity to articulate their ideas in ways that feel meaningful. These submission methods are powerful - a natural extension of what students are already using outside of the classroom to express themselves and teach their peers online. Video and audio help students showcase their unique perspectives and personalities. They also give teachers a clearer view of what students know (especially as AI tools make it easier to produce written responses). Assessment should go beyond testing knowledge. It should create opportunities for learners to express themselves, build confidence, and take pride in their education. When students have the right tools to share what they know in personal and authentic ways, they gain a deeper connection to their learning - and a clearer sense of themselves.

3 ideas from the report "Assessment Reform for the Age of Artificial Intelligence" The emergence of generative artificial intelligence (AI), while creating new possibilities for learning and teaching, has exacerbated existing assessment challenges within higher education. However, there is considerable expertise, based on evidence, theory and practice, about how to design assessment for a digital world, which includes artificial intelligence. AI is not new, after all, even if the current iterations of generative AI are. This document, constructed through expert collaboration, draws on this body of knowledge and outlines directions for the future of assessment. It seeks to provide guidance for the sector on ways assessment practices can take advantage of the opportunities, and manage the risks, of AI, specifically generative AI. 3 key ideas: a.Multiple and Inclusive Assessment Approaches: The document emphasizes the need for diverse assessment types to form trustworthy judgments about student learning in the context of AI. It advocates for triangulating different assessment methods to enhance inclusivity and reliability, recognizing that no single assessment can capture all aspects of student performance in an AI-influenced environment . b.Integration of Continuous Feedback and Reflection: The examples provided illustrate a shift towards breaking assessments into smaller, manageable parts that allow for ongoing feedback and reflection. For instance, in a Bachelor of Science program, students are encouraged to document their decision-making processes and critical thinking, which helps them learn from their experiences and understand the relevance of their tasks . c.Long-term Assessment Planning: The document highlights the importance of strategic planning in assessment design across academic programs. It suggests that institutions should review and adapt their assessment strategies over time, integrating multi-method approaches to ensure that key learning outcomes are met while maintaining academic integrity and security .

Most people discuss building with LLMs, but almost no one discusses evaluating them. And that’s exactly where teams go wrong. We don’t just need powerful models; we need reliable, measurable, decision-ready models. I put together a simple visual that breaks down four ways to evaluate LLMs and exactly when to use each method. (Benchmarking 🔍 | Human Evaluation 🧑⚖️ | Automated Metrics 📊 | Adversarial Testing 🐞) Here’s why this matters: ✨ Benchmarking tells you how your model compares ✨ Human evaluation tells you if anyone will actually trust it ✨ Automated metrics tell you if it scales ✨ Adversarial testing tells you if it breaks Together, they paint the real picture of model performance. If you’re working with GenAI, building AI products, running experiments, or optimizing pipelines… your evaluation strategy matters more than your model choice. 🧵 I’ll share a deeper breakdown of each evaluation method soon. Until then, save this post for your next AI project. 👇 Which evaluation method do you rely on the most? Let’s compare notes in the comments. #AI #MachineLearning #LLMs #AIEvaluation #GenAI #DataScience #LLMEngineering #AIProduct #FutureOfWork #TechCommunity