For past few months I have been working on the iOS application for Ziro Fit. This application is a natural extension to the overall web application we have https://ziro.fit/en. In the web version, we had a functionality for marketplace where you can find and filter trainers. It was more focused on client overall. In this case client refers to people who are looking for personal trainers nearby them. As a marketplace, It lets clients find trainers by goals and locations. And trainers had a dashboard, online profile features, booking calendar and more.

After collecting a lot of feedback on how a trainer would like to use an application. We decided to go forward with native iOS application. We could integrate faster animations, real-time haptic feedback, connection with apple health and superior usability.

Web Challenges Overcome

With web version, we had major issue of user experience. The web application was getting over crowded and faced with clunky navigation.

Progressive Web Apps can’t match the native iOS experience.

Cross-platform frameworks were considered but rejected; they often demand near-native effort yet deliver compromises in polish, speed, and native feature depth.

The application focused on calendar/schedule management, exercise templates, a home dashboard, clients list and live workout. Out of which live workout was one of the most important screen. It had exercise logging features. Where personal trainer will spend quite a lot of time. These exercise login features involved selecting of exercises, filling correct data like reps and sets, having rest.

App has two different modes. One is professional mode which is for fitness trainers and other is personal mode which is for clients. Based on the login account, the app detects the respective interface to show.

Trainers see business metrics like earnings and calendars, while clients access workout logging and trainer discovery.

Trainer Mode: Professional Tools for Pros

Let’s look at some of the screens for personal trainers.

Professional Mode> Home

The professional dashboard is designed with business in mind. In this home screen, trainer can see how much money they are making, how many clients they have, how many check-ins they have and how many bookings they have. Here check and refers to status update from existing clients. Client can report on progress of workout , metric like weight or reps and adherence to the plan designed by trainers. Clients can tell how good they slept, how was their nutrition, and overall how active they were.

There are quick few actions for trainers. They can add a new client, can design a new program, set up the storefront, and manage their booking calendar.

Professional Mode> Programs

The program screen is designed for building templates of exercises. It offers tools to build exercise sessions for their clients. The trainers can design session by session, exercise plans for the client. They can sequence exercises, set reps/sets and save reusable templates for efficiency. And you use these templates later for other clients too.

Professional Mode> Calendar

The calendar screen enables scheduling and real-time live session of clients. One of the main view is calendar stripe view where they can tap On the date to load that day’s exercise sessions and client cards. They can see whether that session is ongoing, completed or planned. Completed session appear with green badge. Tapping on the card reveals the full workout history including the exercise performed, rep/sets logged or any notes or metrics from that session. The design also provides quick access to the past performance data for reviewing clients progress or any follow ups. On tapping the plus button, the trainer can schedule a new session. They can set a date, set the exercises sets need to perform.

Professional Mode> Client Screen

The client screen provides the professional overview to manage the clients. It is list of clients arranged in alphabetical manner. Clicking on the client card reveals the previous and upcoming sessions, the growth and progress of the client.

Share Health by Design

Client Mode: Precision and Discovery

Coming to the personal mode. It focuses on interfaces for logging, discovery and motivation for the client.

Personal Mode> Home

It let clients elevate their fitness journey with coach trainer, their program status, weekly check ins, quick start of a workout and recent workouts.

Personal Mode> Explore

The explorer screen serves as a trainer discovery hub where clients can uncover elite trainers and exclusive fitness events. They can browse and filter the trainers based on their ratings, location and categories. For events clients can see how many people are attending, proximity and details of the plan.

Personal Mode> Live Session

The live workouts let clients track sets in real time during sessions with the reps, RPE, weights, timer and controls like pause and cancel.

The application is now available to download on iPhone.

https://apps.apple.com/au/app/ziro-fit-connect-train/id6758610548

Other Features

Additional features include appearance like dark/ light mode and multiple languages. We have introduced Polish for the moment. English, German, Spanish and French are in upcoming version. Explore menu also have an interactive map, not just the list and grid view. Trainers also have profile customisation with their photos, training methodology, sessions, packages, testimonials and reviews.

Upcoming update

In the next upcoming update, we are working to fully integrate the online payment system with Stripe. This addition will enable trainers to handle transactions directly in the app, streamline their earnings, track them on their professional dashboard, and reduce the friction for client bookings. While clients can enjoy instant checkout during trainer discovery. Clients can purchase multi session bundles up front during the discovery phase locking in commitment and providing trainers with predictable revenue streams.

Lastly, we also plan to introduce native Android application in coming months. This will mirror the smooth UX of the iOS version. Live logging, calendar tracking and dual mode. While resolving cross-platform gaps. And reaching vast Android user base.

So here I present you in no particular order or categorisation:

If any of these websites inspired you, I’d love to help bring that same level of thoughtful design and clear branding to your health startup. At Ziro, we’re passionate about creating digital experiences that truly connect with users and make a difference. Reach out anytime, let’s build something amazing together.

https://www.ziro.health/

In the video above I walkthrough the home page of http://ziro.fit/ . Showcasing trainer search, exploring filters, map, trainer profile, packages, reviews, book session and calendar.

The platform is part of Ziro’s broader digital health solutions, which focus on user-centric design to empower individuals and enhance care through intuitive, impactful technology.

Updated Video with more volume

What is this?

Ziro Fit (https://ziro.fit) is a product built for fitness management. Its a fitness management platform. Currently we are targeting fitness trainers and their clients.

It allows fitness trainers to manage schedules, track progress and handle bookings. While clients can easily discover, book and train with verified professionals*

We are building an easy way to organise, track and grow a fitness business.

What it offers?

For Fitness trainers it offers:

• Scheduling and Appointment Management: Easily manage appointments, handle cancellations, and send automated reminders to clients.

• Client Profiles and Progress Tracking: Detailed client profiles enable providers to track progress, set goals, and monitor key metrics

• Enhanced Client Reach and Engagement: Trainers can reach more potential clients, both locally and online, increasing their visibility.

For clients it offers:

• Find trainers: Clients can easily search for trainers based on location, specialisation, and availability.

• Reviews and Ratings: Clients can read reviews, see ratings, and compare trainer profiles, helping them make informed decisions and choose the best fit for their goals.

• Transparent Pricing and Packages: Clients can view pricing options upfront, compare packages, and find plans that suit their budget, reducing uncertainty.

Dive deep into the product itself:

1. Find trainers : Online or offline, speciality and location.

   

2. Filter out trainers: Allowing clients to filter trainers by specific needs such as fitness goals, training styles, or credentials, ensuring better matches.

   

3. Detailed profile of trainers: Client reviews, trainer certifications, and platform provide reassurance of trainer quality and safety

   

4. Plans and Packages offered by trainers:

   

5. Progress Tracking and Support: Tools for tracking workout progress and communication, enhancing motivation and accountability.

   

6. Client management tools:

   

   

What’s now?

The Ziro fit will bring trainers and clients, making personalized fitness coaching more accessible, efficient, and scalable. It presents a compelling opportunity for growth in the digital health and fitness sector.

We are working together to make personalized fitness coaching more accessible, efficient, and scalable. Offering a compelling proposition for users(clients) and fitness professionals(trainers) alike.

Feedback is welcome. Trainers are invited to connect and share their insights to help improve the platform together. Join the conversation and shape the future of fitness coaching.

Ziro Fit is a product of Ziro

Our goal is to help digital health innovators grow faster through design that builds trust, adoption, and lasting impact.

https://www.ziro.health/

This time I had an idea to test some new stuff. So, I opened up Gemini for image generation on the side tab, to test its nano banana - the image generation model.

In this article you will see what I tested by combining multiple images and prompting to produce worthy compositions.

Nano Banana : A game changer for visual creativity

Nano Banana is the most recent model of image generation from Google Gemini. It allows to edit images using natural languages with consistency, details and photorealism. Pose changes, background swaps, objects additions among many other features. Making it ideal for industrial design compositions. It allows images to edit multiple times with complex editing.

Let’s check what the hype is all about. In the next stage I wanted to test out the tool, without no over expectations.

Collecting References

I started collecting images of different products not related to each other. A shoe, a earphone, a bike and an iPhone.

Editing with Nano Banana

Prompt 1 :

Athletic woman in black sportswear and red-tinted sunglasses, standing confidently with a sleek, modern look against a plain background.

I attached the image of shoes and earphones, there was no reference image of woman attached.

I was already in awe and inspired with the first image.

Prompt 2 :

Add bike

I attached cropped image of the bike.

Prompt 3 :

Add iPhone

I attached image of the orange iPhone.

Prompt 4 :

Now at this point I wanted to stop, not making the composition over crowded. But I wanted to change the pose. What the maximum it can do?

Make her breakdance

This was absurd and crazy. I was beyond impressed. It’s bananas. It truly lives up it is name.

The final image was composed in stepwise process. Without dumping images in one go. Making incremental changes.

Share

To my further investigation I found a tool (not tested yet) which can help in making variations systematically and keep style alignment. https://www.kittl.com/features/ai/flows

This is also in line with my previous thoughts on node based ai editors. Please check that article here:

With Nano written in the name of the model, I can safely assume there is a bigger model cooking in Google’s labs.

Conclusion

We can conclude that:

1. Can make great believable images.

2. Can maintain character with previous details in addition to new objects.

3. Can work with incomplete/cropped references.

4. Can change poses of the characters. Even dramatic poses.

In my search I didn’t find much information pertaining industrial design and nano banana. I am pretty confident, this will change change my composition workflow.

In following weeks I will experiment with material & style, integrate with environment and scene composition.

Announcement:

I'm hosting free branding workshops for healthtech startups and companies wanting to improve their market presence. This two-part sessions will help participants reassess their current branding approaches. Book your session here:

https://calendly.com/ziro-nikhil/30min

For One Page Landing needs check my work here:

https://ziro.agency/

For health related design work here:

https://ziro.health/

This article discuss my current workflow, new ai based design tools and future of the product design itself.

Practical usage in my everyday work

I work with one page no code landing pages with https://ziro.agency/. With goal of creating high converting one page website done and delivered in short amount of time.

My design stack includes Google Forms, Perplexity and Framer.

1. Google Forms: In this step I collect the requirements from my clients. I ask my clients to fill predefined questionnaire. These question detail out the expectations of the clients. What they want to achieve with their websites? Which websites they like currently? Mood and feel….

2. Perplexity: In this step I take text from google forms to generate a copy of landing page. Based on the need I adjust my text prompt in perplexity. Whether a clients high converting page or a contact form or a promotion of their product. This steps requires a lot of iterations to achieve a workable copy for the website.

3. Framer: In this stage I work on design for the website. Going from section to section. For example hero section, who we are, service section, contact form and more. Also to mention I collect lots of references and examples before starting this stage.

I have used this methodology for multiple projects now and it has given me great results.

Website generator Tools : Replit, Lovable

I worked with tools like Replit and Lovable to generate entire landing pages. In my observation in current states the results can be used as generic website but not as well designed products. The output lacks character, tone and personality. The results need to be adjusted and refined.

Repit and Lovable in current stage are not great to be used as “design tool”. Other thing is which I don’t appreciate is endless prompting. More like doom prompting. To get good results visually is quite hard.

Rise of Node based AI platforms

It is clear to me now that one single AI tool can be answer to solve difficult design challenges. It will be series of ai tools that combine together to form final result. The answer is Node based tools.

What is a Node based tool?

A node can be considered as a unit which is responsible to perform a specific/specialised task. These units can be combined together like wires to each other to form a graph. It is a visual process in which one can easily understand how data is flowing from one to another.

This approach has been around ages in 3d softwares. For example, Rhino 3d is a popular 3d design tool for designers. Rhino 3d offers a feature called Grasshopper

https://www.rhino3d.com/

AI Node Tool : Weavy

One such tool I found is called Weavy (https://www.weavy.ai/). In the image you can see how a 3d model of a bust was converted into a video, by controlling each of creation at minute level:

1. 3d model was generated using Rodin(Ai tool for generating 3d models).

2. Combing 3d with color reference to generated an image using stable diffusion.

3. Text prompt and another reference image to generate the final video using minimax.

https://www.weavy.ai/

Share

Future of product design

AI tools has enable to produce products faster than a product designer can catch up. It’s Product First Design Later.

Envision a future where product are build at speed of the idea itself. Followed by iterating and refining product by talking to actual customers. Design in traditional UI/UX way will take a backseat. In my opinion, this is how new way of doing digital products:

1. Idea for new business/product. Plain rational and creative thinking.

2. Prompt, iterate and build quick prototypes. Transfer it to developers.

3. Start testing will user, collect feedback, improve UI/UX and repeat. Validate Value proposition

Idea > Build, Validate and Iterate > Improve Experience with design

Takeaways

As a digital designer it has become prominent that AI has changed the way we work. Product first design later and node based methodology are here to stay.

Special Announcement:

I'm hosting free branding workshops for healthtech startups and companies wanting to improve their market presence. This two-part sessions will help participants reassess their current branding approaches. Book your session here:

https://calendly.com/ziro-nikhil/30min

This is the vision presented in the research paper titled “AI nutrition recommendation using a deep generative model and ChatGPT1” by Ilias Papastratis et al.

Why do we need such a system?

The paper seeks to address concerns of AI generated nutritional advice (accuracy and trustworthiness) by combining it with expert knowledge of nutritional guidelines. Doing such will produce accurate, personalised and nutritious meal plans.

How does this system works?

The proposed system works in three major stages.

In first stage, it uses VAE(Variational Auto Encoder) to model user’s health data(eg. age, weight, activity level etc.) and maps it to latent space. Followed by a RNN(Recurrent Neural Networks) to generate weekly meal plans.

In second stage it employs novel loss function and optimiser. The purpose of loss function is to align the network with nutritional guidelines(EFSA and WHO). Followed by an optimiser to adjust meal quantities for the user.

In third stage, ChatGPT is introduced. The purpose of ChatGPT is to expand the meal variety and generalisation.

The advantage of using ChatGPT is its speed and infinite pool of meals from other cuisines. It can overcome the limitations of databases in traditional systems.

Role of Loss Functions

The loss functions help the RNNs in getting accurate and personalised recommendations. They penalises the network if it deviates from guidelines defined by EFSA and WHO. There are four different loss functions:

• Macronutrient Penalty Loss (Lmacro) : Helps in minimising difference between predicted and recommended ranges for micronutrients.

• Energy Intake Loss (LEI): Helps in meeting user’s energy requirements.

• Kullback–Leibler Divergence (LKLD): Trains encoder for effective sampling and generalisation of meal plans.

• Cross Entropy Loss (LMC): Helps meal classifier to accurately predicts personalised meals.

The total of above helps structure the latent space based on user similarities and differences.

Role of Optimizer

The optimizer makes sure that each meal portion gets targeted energy intake(EI). It calculates the difference predicted values and targeted values. And then adjusts meal portion to minimise the difference.

Enhancing the meal variety using ChatGPT

Till now the meal generated were coming from the limited database of nutritional guidelines. The use of ChatGPT at this stage expands the meal database. ChatGPT helps in generating meals considering equivalent energy intake and macronutrients from other cuisines. Further optimiser helps in achieving 0.00% calorie difference and 12% average accuracy improvement for macronutrients comparing to ChatGPT itself. Lastly it can be adapted for specific dietary needs like allergies and preferences.

Validations on Real User Profiles

Evaluation was done on 1000 real user profiles from two datasets. The system achieved 100% on average energy intake and 84.19% for average macronutrient accuracy for all user groups.

In Summary

The proposed system achieved:

1. Accuracy: Balances AI efficiency with rule-based nutritional science to avoid generic recommendations.

2. Adaptability: Handles diverse dietary needs (e.g., allergies, preferences) through real-time adjustments.

3. Trustworthiness: Validated via experiments with both synthetic and real-world user data, demonstrating scalability.

Takeaways and potential applications

Such a system can be integrated in health dashboards, fitness apps and dietary consulting services. The recommendations will be highly valued. Such a system can become critical part of AI Health agents. I have discussed on it in detail here:

And check out the paper:

This is an ongoing article, would appreciate any suggestions regarding mistakes or comments. Your insights will be invaluable in enhancing the quality of my work.

Healthcare is undergoing a revolution, and I believe design is at the heart of this transformation. At Ziro, we’re not just building products—we’re shaping the future of healthcare. By combining thoughtful design and cutting-edge technology, we help companies and startups bring visionary ideas to life, creating solutions that truly make a difference.

Ziro's Philosophy: The Art of Nearness

Our guiding principle, "Nearness Implies Separation," reflects our commitment to innovation in digital health. We believe that by balancing proximity and distance—collaborating closely with clients and users while maintaining a critical perspective—we can challenge assumptions and create solutions that integrate seamlessly into the lives of patients and healthcare providers.

What We Do

At Ziro, we focus on crafting impactful solutions that address real-world challenges in healthcare. Here’s how we do it:

• UI/UX Design: Designing seamless, human-centered experiences that empower users and drive engagement.

• Branding & Identity: Building bold, memorable brands that stand out in the healthcare space.

• Digital Product Development: Developing innovative health solutions and apps tailored to unique needs.

• Product Packaging: Creating designs that leave a lasting impression—because first impressions matter.

Why It Matters

Healthcare is deeply personal, yet it often feels disconnected from the people it serves. At Ziro, we’re driven by a simple but powerful idea: healthcare should work for humans, not the other way around. Design is our tool to humanize healthcare—whether it’s simplifying complex systems for clinicians or creating intuitive apps that patients actually want to use. We believe that innovation happens where empathy meets technology. My journey—from architecture and industrial design to digital health—has taught me that solving big problems requires a holistic approach: one that considers not just functionality but also how people feel when they interact with a product or service. This philosophy is at the core of everything we do at Ziro.

A Journey Toward Perfection

We view perfection as an ongoing journey rather than a fixed destination. Our iterative design processes are designed to adapt to evolving user needs and technologies, allowing us to strive for "close to perfect" solutions. This philosophy emphasizes that "perfection in proximity" is essential for transforming digital health and enhancing patient care and engagement.

Share

Let’s Build the Future Together

Based in Kraków, Poland, Ziro thrives on collaboration with visionaries who share our passion for better healthcare. Whether you’re a startup with a bold idea or an established company looking to innovate, I’d love to connect and explore how we can create something extraordinary together. Visit 👉 ziro.space to learn more about what we do.

About Me

I’m Nikhil Singh, a digital health entrepreneur and the founder and CEO of Ziro. My journey has always been about pushing boundaries—merging design and technology—to create solutions that truly improve lives. Before founding Ziro, I served as Chief Design Officer at Nutrix, where I helped develop groundbreaking health products. Together, let’s reimagine what’s possible in healthcare—with thoughtful design leading the way.

#DigitalHealth #UXDesign #HealthcareInnovation #HumanCenteredDesign #FutureofHealth

Imagine a future in which you can access a personal skincare consultant that empowers you to confidently select products that align with your skin goals. This is the vision presented in the research paper titled “Deep learning-­ based skin care product recommendation: A focus on cosmetic ingredient analysis and facial skin conditions1” by Lee J et al. The researchers proposes a novel deep learning-based cosmetic product recommendation system. The system combines two key elements: AI skin analysis and Cosmetic ingredient analysis. AI skin analysis facial images for skin concerns and their severity. While cosmetic ingredient analysis predicts the efficacy of skin care product based on composition and order of ingredients.

Lastly is the recommendation module which matches user skin analysis with right products based on ingredient analysis.

The need for such a system

Finding right skincare products can be endless cycle of trial and errors for consumers. Traditional product recommendations have heavily relied on user reviews, ratings and demographics, which can be misleading or not relevant to one’s needs. By analysing the actual ingredients, consumers can understand what constitute in those cosmetics products. Allowing consumers to understand the effect of the ingredients inside beyond marketing tags. Also, this goes well in line by growing consumer interest in understanding how those cosmetic ingredients affect their skin health. Consumer behaviour is going in broader shift towards increased interest in ingredient transparency and demand for personalised experiences.

By combining this AI, it can be possible to take the guesswork out of this process and get matched with products that produce best results.

Skin Image Analyser

The system uses a deep learning model based on U-Net architecture to segment and analyze images of the user’s face. The advantage of U-Net architecture is as it offers accurate segmentation of skin conditions under various shooting conditions. The model has been trained to identify six key skin concerns : pores, acne, wrinkles, pigmentation, redness and dark circles. Estimating grades with severity of each concern.

The model’s performance for skin conditions was validated against expert dermatologist evaluations. Where it showed strong correlations.

Ingredient Analysis

The ingredient analyzer evaluates the potential effects of a cosmetic product based on its list of ingredients. It utilises deep learning model called transformer to analyze the ingredients. It treats the list as a sequential data where order of ingredients matters. The regulation in most countries require ingredients to be listed in descending order of their quantity. Followed by tokenization and vectorization of ingredient list. Each ingredient is assigned unique index to differentiate from each other. Transformer encoder analyses relationship between ingredients in sequence to predict the effects of the product. Then the output of the transformer is fed to multi-label classifier which predicts the product’s multiple effects.

Report Card for your Skin : Recommendation module

The recommendation module acts as a matchmaker. It combines results from two analyzers - the skin analyzer and the ingredient analyzer. The scores from both analyzers are represented as vectors. It calculates the cosine similarity between user’s skin vector and the efficacy product vector of each product. If the angle between two vectors is indicating value closer to 1, it means high similarity. That product closely aligns with user’s need. Further, the products are filtered based on user’s skin type - oiliness, dryness, and sensitivity. Ensuring product recommendations are safe and effective.

HBD Note: By combining AI powered skin analysis with sophisticated ingredient analysis, the system offers a report card for skin health. The users are empowered with this information to make individual decisions based on their needs. Such system can be offered as an app service as personal skincare guide.

Concerns and Challenges

The biggest concern with current model of ingredient analyser is its data imbalances. Which can impact the effectiveness of the analysis. It was found that accuracy of product categories with fewer sample was lower. Leading to less accurate results of certain conditions like acne.

Second is the marketing claims mentioned on products. Relying on marketing tags might result in exaggerated or misleading results.

The researchers mentioned to continue to gather more comprehensive datasets which in turn will help in enhancing the performance and completeness of the recommendation system. It has potential to become more powerful tool for consumers by incorporating actual clinical trial data. Thus producing more objective and scientifically -grounded assessment of product effectiveness.

The researchers also mentioned to include additional attributes like skin firmness, radiance, cancer and texture in the future.

Conclusion and Takeaways

This paper presents a compelling take on how future of skincare might look like. We can expect future where AI continues to evolve to create more innovative solutions. Empowering consumers to make informed decisions about their skincare journey based on their individual needs. Moreover, we can expect that these system will enable consumers to track their skin’s progress over time, predict future skin concerns and develop hyper personalised products.

The article sits at convergence of several key trends which includes growing consumer demand for personalized experience, transformer models and computer vision models to analyse both skin condition and product efficacy. With a level of detail which was not possible before. The system signifies a paradigm shift towards a more data driven and scientific approach.

Audio summary: Get Matched! AI Skincare Finds Your Perfect Products

In my previous post, I talked about how health agents/longevity agents can become part of everyday healthcare. And here I present you an app concept powered by a longevity agent.

If you haven’t checked it out, check now :

Why SuperNormal as a name for the app?

You must be wondering why I went with such a name.

Become the best version physically and mentally is the “Super” part. While doing things you will normally do on day to day basis. Hence “Normal” part. The daily activities which are guided by an AI technology. The agent understands your behaviour and lifestyle, it decides your progress path which feel natural, progressive and achievable.

Intended User

The intended user for such an app are “Rejuvenation Athletes”. These are individuals who are actively trying to do activities that can slow down the biological aging and help improve their wellbeing. These athletes check their biomarker for aging like DunedinPACE(novel blood biomarker of the pace of aging). The concept itself has gained popularity as of initiatives like Rejuvenation Olympics where individual try to achieve best score for slowing of aging for leaderboards.

The Rejuvenation athletes focus on number of health interventions which can help them reduce aging process. This include diet, exercise, sleep, stress and medical therapies. One of the popular example is of Bryan Johnson who is documenting his process and calling it “Blueprint”.

The Concept for the application

The app brings ideas from various themes - aging clock/biomarkers, multiple connected monitoring devices, the community revolving around rejuvenation athletes, and lastly the advice from AI agents.

The Information Architecture

I divided the information architecture in two levels. The first level caters to three distinct purposes - Aging/Biomarker related content, community engagement and personal progress/statistics.

With each having their own uniques tabs in the application. The second level is catering to additional information to support level one.

The Color and Font System

I choose an overall dark theme for this app, which gives it more premium and tech-savvy vibe. The color scheme is dynamic and contrasting in nature. Making it suitable to attract attention. Overall colors are bright in nature to contrast with background black. Which makes the design of the app very modern.

The Home Screen

The homes screen brings vital information relevant to a rejuvenation athlete. Here user gets information about his aging score. The primary one is intended to be collected from an invasive blood test - an aging test. Amount and frequency of such a test needs the help of a medical practitioner. Second information is for cumulative score which is generated by how good an individual has been doing their aging protocols and interventions. Followed by recommendations from a health agent. Later in post I discuss, how such a screen might look like.

The Leaderboard

The leaderboard gives glimpse of individual’s ranking compared to other athletes. The board is inspired by an actual leaderboard for rejuvenation athletes. Maintained by Rejuvenation Olympics.

The User Progress

The last tab will be dedicated to user progress. The user can select the health parameter like sleep, nutrition, exercise etc. Which will be followed by detailed graphs and charts relevant to that parameter.

User can select any given parameter and try to compare with a competing athlete. Given the fact this can be sensitive health data, it might be necessary to anonymise or take relevant permission from other party. Following is possible idea, which is based on readily available data from aging leaderboards:

The Health Agent

The health agent which continuously improvises and reads on your health data will prepare a list of actionable items. These items will have sole goal on improving aging and well being of the user. Day by day action items which user needs to take. In current form the idea is very linear. Which I believe can be improved to have more dimension to it. Making it more interaction and inviting in nature.

The Community and the Gamification

What I have realised that “Rejuvenation Athletes” are already competitive in nature. They are working to make best out of them. I propose challenges which can be used for a healthy competition among themselves.

A badge system for showcasing achievements. These badges are awarded to the user once they complete the challenge. The challenges are designed in a manner that it represents aspects of longevity and fitness overall. For example “Sleep Warrior” will be given to someone who completes 8 hour sleeping challenge for 30 days straight. The given badges are generated using PicLumen Realistic V2.

Once the user completes a challenge, they are rewarded by the app. This will foster a sense of competitiveness among users.

Conclusion

Supernormal presents here a compelling idea on how a longevity app can look like. It is powered by a health agent which gives insight and directions for a rejuvenation athlete. Aiming to play a vital role in journey of such an athlete.

This is an ongoing article, would appreciate any suggestions regarding mistakes or comments. Your insights will be invaluable in enhancing the quality of my work.

This article is written by Nikhil Singh. He is a digital health entrepreneur and the founder and CEO of Ziro, a company focused on designing next-generation healthcare experiences and user-centric digital health solutions.

Imagine a world where your health is managed not by infrequent doctor visits but through a personalised digital advisor that monitors your well being 24/7. This is the vision presented in the research paper "AI Health Agents: Pathway2vec, ReflectE, Category Theory, and Longevity,1" by Melanie Swan and her colleagues. They introduce the concept of AI Health Agents—personalised digital advisors designed to facilitate "healthcare by app" rather than "sickcare by appointment."

It is expected there will more than 2 billion people over 65 by 2050. So need of such innovation has never been more pressing.

The Longevity Mindset

The paper proposes change in mindset when we think about longevity. First to treat aging as a disease, instead of an inevitable condition. A condition which can be treatable with potential interventions to slow, reverse and decline the process itself.

Second is to be pro-longevity tech. Meaning be welcoming in accepting and building technology for longevity like AI health agents and personalised aging clocks. Not only to individuals but large scale too.

The Power of Continuous Monitoring

Longevity agents or just health agents are future of personalised care. They are AI algorithms designed to take care of you. Focusing on maximising your lifespan(long life) and health-span(quality life). They continuous monitor parameters about your health. Parameters which are measured by number of devices for example smart watches, smart rings etc. Such interfaces communicate with you in natural language via app interfaces. And can take decision which is in best in interest of you.

On a population level they can help monitor thousands of patient simultaneously. They can reach individuals which are separated by the digital divide. Ensuring equitable access to these technologies will be crucial for transforming into a reality for all.

A future in which Precision Medicine as a service will be the norm.

The Science Behind AI Health Agents

These longevity agents have several different components. Which come together to make intelligent decisions. While continuous monitoring is one, the other is maintaining a digital twin of you.

A digital twin is a concept where physical entity or process is presented with virtual copy. In healthcare - a biomedical digital twin. They can be used to run simulations, education, research and drug development.

They combine and integrate data from various sources - such as biomarkers and aging clocks. Can find patterns which are hard to spot by traditional methods. Thus paving a way for precision medicine. Which can be tailored to individual needs.

The aging clock is a concept which compares biological age to chronological age. Thus measuring our rate of aging. Can possibly identify areas of interventions. They can analyse various factors like epigenetic, transcriptomics, glycan, metabolic and telomere length.

These agents would utilise generative AI, graph neural networks and knowledge graph embeddings to process vast amount of biological data and create comprehensive models of you.

Digital Biology and AI

By bringing Gen-AI methods to biology, it might become possible to explain biology via mathematics. Which hasn’t been the case till now. Pointing a future when AI can derive mathematical formalisation of biological processes.

Comparing to language models, biological AI models may be orders of magnitude complex. Examples of such models are Protein and Genome language models.

Blockchain for Securing Health Data

For storing and managing health data, it is proposed to use blockchain. The paper discuss projects like MediLedger and Triall. Demonstrating real world case scenarios of the application of blockchain technology in biology and healthcare.

Due to decentralised and encrypted nature, data on blockchain is inherently secure. Making it suitable for storing sensitive health data.

Biocurrency and Bio Wallet

The paper proposes a new system for tracking health parameters and interventions using blockchain. Representing biological entities like organs, biomarkers or interventions as “biowallets” with “biocurrency” balances. These biocurrencies maintain homeostasis. Homeostasis is process by which an organism maintains stable conditions and adjust it to best suit its survival. These biowallets will hold balance of the current state of biomarkers.

By tracking changes of biocurrency, it will be possible to constantly refine the protocols and projected outcomes.

Pathway2vec and Personalised Longevity Protocols

The paper introduces a method called “Pathway2vec”. It represents biological pathways vectors for ANNs(Artificial Neural Networks). Simply, biological information into numerical information so that AI systems can analyse. With goal to create personalised longevity interventions - Personalized Aging Clocks. This method is similar to word2vec which analyses language. It can also be used to represent genes, genetic variations, mutations and diseases. With core idea to slow down aging process and develop tailored personalised longevity protocols.

Risks and limitations of Health Agents

The most important will be how fast we can build such AI systems. It might be too late given the aging population. Other major risk is privacy and security of sensitive health data. How it gets processed and implemented by AI systems.

Conclusion and Takeaways

The paper presents a compelling take on how health agents can become part of everyday healthcare where personalised advice and continuous health management becomes the norm. With potential to significantly improve human lifespan and overall wellbeing of an individual.

Given above was a detailed interpretation of the article AI Health Agents: Longevity, Pathway2vec, ReflectE, and Category Theory. The article sits at convergence of AI, digital biology and blockchain technology.

This is an ongoing article, would appreciate any suggestions regarding mistakes or comments. Your insights will be invaluable in enhancing the quality of my work.