EnduCloud

Home
Ireland
Dublin
EnduCloud

WhatsApp: +972 55 291 5573 EnduCloud monitors the response to training and modifies the plan based on the individual response.

EnduCloud, a digital training and health platform built on advanced data analysis and artificial intelligence, designed to support athletes, recreational trainees, and professionals in making more accurate decisions over time. Prescription, analysis and individualization of physical activity for health and performance

EnduCloud assist health providers and coaches in the prescription of aerobic physical activity plan for health and performance based on the individual targets and constrains. EnduCloud combines advanced (and unique) advanced analytical tools and artificial intelligence to analyze the training load, the response to the load and the individualized adaptation. EnduCloud uses the wide range of wearable devices that are already in use by the people from sport and smart watches with training time or 24*7 monitoring, which is collected through common applications including Strava™ Garmin Connect™ and Polar Flow™, and will gradually add more vendors and sensor types. EnduCloud include specific modules customized to the needs of specific populations starting with cardiac patients and continuing to general health, recreational athletes and oncological patients. At its initial stage EnduCloud supports walking and running, that will be gradually expand to include cycling, swimming, triathlon, rowing, cross country ski and more for indoor and outdoor sport. EnduCloud primary audience is health providers and coaches that uses EnduCloud to provide higher quality personalized service to their customers. Additionally, EnduCloud offers a free tier of unsupervised personalized service to the general public, that can be complemented by paid supervision services through EnduCloud. EnduCloud elaborate database which includes comprehensive information regarding the trainee, the training plan and the trainee response to the plan will use both for the next stages of EnduCloud’s development and to improve our understanding on the consequences of physical activity.

31/10/2026

When two training-load metrics agree, are they measuring the same thing, or just the same clock?

A methodology review published this year in IJSNEM (Bossi, Matta, Lima & Bara-Filho, 2026) walks through one cyclist's 76 sessions and three common load metrics: TSS from power, Edwards' TRIMP from heart rate, and session RPE (how hard it felt × minutes).

On the raw numbers they look interchangeable: session RPE predicts TSS with R² 0.82. Then the authors remove session duration from both sides. What is left: session RPE vs TSS, r = 0.59. TRIMP vs TSS, 0.29. Session RPE vs TRIMP, 0.22. Most of the apparent agreement was the shared clock: a long session scores high on every metric.

It is one athlete, and the authors say plainly that the equations don't generalise. The method does: before you trust two load numbers because they agree, take duration out. But this may go much deeper - today we trust many of the load matrix because one is correlated to the other, going back to Banister TRIMP from the 1970’s. But if the correlation is based on the underlaying distance, what does it say about the actual load?

This is the problem that STRSS™, the stress marker we are developing at EnduCloud, come to solve. Load is the work you did; every minute adds to it. Stress is how close that work took you to your limit. STRSS is designed to measure the second:
• 100 means a maximal effort for you, whether it lasted 5 minutes or 3 hours. Above 100, you went past your own curve.
• A long easy session stays well under 100, however many hours it ran.
• The same effort scores the same from power, pace or heart rate, in any endurance sport.
• It is anchored to each athlete's own capacity curve, not to a 60-minute convention.

To be clear about where we are: the version in the app today gets part of the way. It is individual to each athlete and agrees across power, pace and heart rate, but long easy sessions still add up more than they should. Closing that gap is the next version, and this paper gives us part of the test. Our own check of heart-rate vs power agreement is inflated by shared duration too, so we will judge the next version with duration taken out.

26/10/2026

A number on your wrist is a measurement, not the truth

Last week one of our athletes went for an easy, steady jog. For about 13 minutes his watch recorded a heart rate of around 155 bpm, peaking at 160. Then, in 19 seconds, at the same pace and cadence, it fell to 118 and stayed there. Hearts don't do that. Optical wrist sensors sometimes do.

The cost wasn't the bad number. It was everything built on top of it. The watch showed the user zone 3–4 on a zone 2 session, so he slowed down to fix a problem that didn't exist. The same number would also have inflated his training load, scored his ex*****on, and could have become his new "max heart rate".

There's an obvious shortcut here: export the session, paste it into an AI chat, ask how it went. A language model given this run sees 160 bpm and has no reason to doubt it. Models are very good at explaining whatever they're handed, so it will tend to explain the number (heat, fatigue, a dip in fitness) rather than question the sensor. The answer is fluent, plausible, and built on a reading that wasn't a heartbeat.

We work in the opposite order: physiology first, then the model. Deterministic checks look for what a heart physically can't do, like a 37 bpm drop at a steady effort, or an optical sensor locking onto the step rate. Those stretches are masked, never deleted, and kept out of training load, ex*****on scoring and heart-rate thresholds. The AI coach then gets the clean data plus the verdict: what happened, which numbers not to quote, and to judge the session on pace, duration and the heart rate that held up. We give the model everything we know, including what we know to be wrong.

The athlete is told too. Their chart shades the stretch and explains it. And we deliberately didn't switch their targets to pace: they chose how to train, and a sensor fault is ours to name, not theirs to be re-planned around.

Now rolling out in EnduCloud on runs and rides, with older sessions being re-checked next.

21/10/2026

The American College of Sports Medicine has published its 2027 Worldwide Fitness Trends.

EnduCloud was built on a few assumptions: that the watch would become the center of how people train, that the coach would live in the phone, that one engine should serve a marathoner and a 70-year-old in a community group, each at their own level, and that strength, yoga and Pilates belong inside an endurance plan rather than next to it. This year's list reads a lot like the brief we wrote for ourselves. Seven of the top ten are things we build. Wearable technology is No. 1 for the 10th time. Mobile exercise apps climbed to No. 2, and balance, flow and core strength to No. 3. Fitness programs for older adults are in the top five for the fifth year running, and traditional strength training rose to No. 5.

The one that hasn't moved is the one we care about most. Data-driven training technology ranked No. 8 for the second year in a row. ACSM defines it as using HRV, sleep, training load and estimated VO2max to fit training to readiness and recovery. Almost everyone collects the signal; far fewer turn it into tomorrow's session. Athletic trainers, who make readiness and return-to-play calls every day, ranked it No. 2.

Closing the loop between knowing how you are and knowing what to do is the whole point of EnduCloud. Every morning it combines sleep, HRV, resting heart rate, recent training and how you say you feel, and adjusts the plan, with the reason for every change. Your Tuesday Pilates class is part of that picture, not a surprise to it. Where there's a coach, the coach stays in charge: larger changes wait for their approval. And it works with the watch you already own: Garmin, Suunto, Polar, COROS, and Apple Watch through Apple Health (iPhone app in beta), with Oura and WHOOP built and next to switch on.

It's live, used daily, and still early. We take the paper's own caution seriously: focus on metrics that matter, and don't over-read estimates whose accuracy varies from device to device. The survey itself measures how widely professionals see a trend, not how well it works.

19/10/2026

Exercise is a prescription. That is the frame of the American College of Cardiology's new 2026 scientific statement, and it is overdue.

The statement starts from an uncomfortable fact. Low cardiorespiratory fitness predicts outcomes as strongly as smoking, hypertension, high cholesterol or diabetes. Yet most clinicians get little training in prescribing exercise, and most adults never reach the targets.

What it asks for:
• Prescribe it like a drug: frequency, intensity, time, type and progression, fitted to the person's fitness, goals, symptoms and medications.
• 150–300 min/week moderate, or 75–150 vigorous. The biggest gain comes from going from none to some. Activity packed into 1–2 days is linked to similar benefit.
• Strength training twice a week is now a core component, not an add-on. Sitting time carries risk of its own.
• About 7,000 steps a day is linked to a 47% lower risk of death compared with 2,000.
• Most people without symptoms don't need a routine stress test before starting. New symptoms during exercise mean stop and see a doctor.

The section I keep rereading is the one on wearables. They add about 1,850 steps a day, and a third of users abandon them within six months. In a meta-regression, only two components independently predicted the effect: personalization and text messaging. The number of features did not.

That is the bet we've made at EnduCloud. Plans are built around your goal, your fitness and the days you can train, and every change comes with its reason. You see your weekly activity minutes against the WHO range, strength sits inside the plan, and each morning you get a read on sleep, HRV and resting heart rate. It works with the watch you already have. It is a training tool, not a medical device. For anyone living with heart disease, the prescription belongs to their medical team; our job is the layer between visits.

Source: Myers et al., JACC 2026, doi:10.1016/j.jacc.2026.09.008

13/10/2026

The Top 20 Semi-Finalists of Go Healthy with Taiwan 2026 Have Been Announced! 🌍

This year’s competition received 832 proposals from 62 countries, setting a new record since the program began!
Following a thorough evaluation, 20 teams from 16 countries were selected as semi-finalists, with Israel proudly represented by EnduCloud Ltd.

👏 Congratulations to EnduCloud Ltd. and all the selected teams! Wishing everyone the best of luck in the next stage of the competition. We look forward to seeing the innovative solutions they bring to the global health and wellness sector.

10/10/2026

Physical inactivity among adults worldwide rose from 23.4% in 2000 to 31.3% in 2022. Over those same years, tracking our own activity became ordinary.

A Current Opinion paper published last month in Sports Medicine, from the Global Alliance for the Promotion of Physical Activity (Steinacker, van Mechelen, Engebretsen, Verhagen, Pitsiladis and nearly 150 authors in all), proposes "precision prevention" with physical activity as its central, modifiable anchor behaviour. Omics, AI, digital twins and wearables are all in there. The sentence I keep returning to is in the conclusions: "technological and biological advances alone are insufficient unless they enable sustained behaviour change."

The authors take an established behaviour-change model (information, motivation, behavioural skills) and add person-generated health data as a two-way feedback loop: what the watch, the app and the diary record should flow back into what a person knows, wants and is able to do. Measurement is an input to the loop, not its outcome. They are equally direct about context: activity trackers do little where there is nowhere safe to walk.

Two more points worth taking from it. Clinicians are asked to assess physical activity as a vital sign and to give written, individualised prescriptions. And the evidence they cite for digital tools is modest but real: an umbrella review of 507 RCTs found an average of about 1,300 extra steps a day and 55 more minutes of moderate-to-vigorous activity a week.

To be fair to the paper, it is a conceptual framework, not a trial, and the authors say so plainly.

It is also the problem we spend our days on at EnduCloud: the distance between knowing how you are and knowing what to do next. Most of that distance turns out to be the loop, not the data.

27/09/2026

A vascular surgeon said something recently that I keep coming back to. Want to improve your health? Start by putting on your shoes. Not a run, not a program. Shoes. Start small and build from there, but whatever you do, don't give up.
On a podcast I listened to over the weekend, Dr. Lily Johnston looked back on five years as a surgeon, and her point was that the wins that really matter don't happen in the operating room. They happen on ordinary days, when nobody's watching, and they start with believing that change is possible. Just as important: surgery fixes one blockage, not the disease that has been quietly building for years, so change needs to start decades before symptoms show up.
That's why, for users who aren't athletes, EnduCloud puts the focus on the days between appointments:
• A daily activity goal, alongside the World Health Organization's recommendation
• A reminder at midday and in the evening if you haven't hit your goal yet (if you opt in)
• A mix of aerobic activity and strength training
• Weight and nutrition tracking
• Working with a coach or a therapist? They get your week and can dig into the details

Bottom line, three things do more for your health than anything else, adding years to your life and, more importantly, health to your years: a whole, healthy diet, physical activity, and sleep.

On the daily activity card, tap "Personalize."

13/09/2026

Most of what we were taught about lactate is wrong. Not "outdated nuance" wrong. Backwards.

Lactate was never the waste product. It is a fuel and a signalling molecule, shuttled between muscle fibres and out to the heart, the liver and the brain, oxidised almost as fast as it appears. It is produced continuously, including at rest and at fully aerobic intensities. And it does not cause the burn: acidosis comes from hydrogen ions released elsewhere in the ATP cycle, while lactate production actually consumes protons. A rising blood lactate reading is not a switch flipping from aerobic to anaerobic. It is the balance between appearance and disposal moving.

And then there is 4 mmol/L. The number a generation of us treated as "the anaerobic threshold" is a population average that belongs to nobody. Maximal lactate steady state has been reported anywhere from roughly 2 to 8 mmol/L in trained athletes. Prescribe to the average and you prescribe the wrong intensity to almost everyone. That is the difference between an easy day that is genuinely easy and one that quietly eats the next three.

We see the same error one layer up, in zone charts. Across our athletes who have both ventilatory thresholds actually measured, the bottom of their "threshold" zone under a generic percentage-based chart sat a median of >10 bpm below their real VT2. Over 10 beats, on every tempo session.

So we anchor intensity to the athlete's own measured thresholds rather than to a fixed concentration or a population percentage, and we convert between zone systems through those anchors instead of copying a table. It is the part of EnduCloud we are least willing to approximate.

The physiology moved on decades ago. Most training software has not.

04/09/2026

One of you is abroad on holiday. One works shifts and can only run at 22:00. One moved to Berlin two years ago. You still want to run together.

Together is now rolling out in the EnduCloud pilot: one event, any number of places. Everyone runs their own route, either at the same moment or whenever it suits them inside a window the organizer sets, and a single live board shows the whole group moving through it.

Three things it is for:
A group of friends, scattered. Four people, four cities, one Sunday long run. You watch each other's progress as it happens, and you all finish the same event.
Organized community events. An event like a national memorial run where thousands take part without a single start line. Each participant runs their own race, at their own hour or all at once, and one public board holds the whole thing together. Sharing a link is the entire registration flow.

Live tracking at a real race. Your family follows you through the course from wherever they are, in a browser, with no account and nothing to install.
Your run counts however you recorded it: watch, phone, file import, health platform. The public board shows progress only, never a map trail or a physiological data, and each participant chooses whether their name appears at all.
Rolling out now on the web and Android and iOS following soon. Early, and we would rather say so.

02/09/2026

We went looking through our own database for a signal we suspected was there, and found something worth sharing, including a part that almost fooled us.
The question: when an athlete is getting sick, digging a hole, or simply not absorbing their training, is there a signal in their training data, before they think to mention it?
We have two numbers for the same session. What the athlete said it felt like (1–10), and what our load model (STRSS™) measured it cost, fitted per athlete against their own critical power and threshold, not a generic formula. Over twelve months, 2,322 sessions carried both, from 73 athletes with enough of each to compare.
The first result was a disappointment that turned useful. Perceived effort is a weak predictor of measured load: across the entire 1–10 scale, load moves only about 1.6×. Duration explains roughly five times more. The reason is session structure, as the median session in every rating bucket was about an hour, so a session rated 9 isn't an hour at 9, it's an hour with hard intervals in it. Effort tracks the peak; load accumulates over the whole.
Which means the difference between the two carries information neither carries alone. When that gap stays elevated across consecutive sessions, we checked it against two measurements that had no part in producing it: resting heart rate ran +0.19 sd, HRV −0.32 sd. Two independent autonomic markers moving the way the hypothesis predicts.
The catch is the most useful thing here. Include ratings below 3 and the resting-HR association flips the wrong way - not noise, arithmetic. On a coarse integer scale a step from 1 to 2 doubles any ratio built on it. A metric derived from a low-resolution input has a floor below which it misleads rather than merely wobbles. If you're building on a 1–10 human rating, find that floor before you trust the output.
Effect sizes of 0.2–0.3 sd from 30 episodes are a nudge, not a diagnosis. So it's rolling out in our pilot as an advisory – supportive input next to HRV, resting HR and sleep. It proposes; it never silently changes a plan. Every flag is stored with its numbers so we can score the calls against what actually happened.
Two of our three findings were negative results. That's exactly why the third is worth anything.

Address

Dublin

Website

http://EnduCloud.com/, https://www.strava.com/clubs/1194592, https://chat.whatsapp.com/L00AfWB1pHBJCEcvMT353A, https://c

Alerts

Be the first to know and let us send you an email when EnduCloud posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.