But what if much of that excitement was manufactured?

Recent disinformation research reveals a disturbing truth: DeepSeek’s meteoric rise was largely orchestrated by thousands of coordinated fake accounts, operating with the precision of a state-sponsored campaign. This isn’t just another case of social media manipulation—it’s a wake-up call for how easily artificial hype can trigger real financial consequences.

From Raw Output to Professional Tables: Transforming Statistical Results with gtsummary

In this article you’ll see the way to communicate to your stakeholders the way to present the results in a clear, accessible format. The gtsummary package transform complex statistical output into publication-ready tables that speak to any audience.

Coefficients:
                    Estimate Std. Error t value Pr(>|t|)    
(Intercept)         45.2341     2.1234   21.29  < 2e-16 ***
treatment_group      8.7654     1.5678    5.59  8.9e-07 ***
age                  0.2345     0.0876    2.68   0.0089 ** 
gender_female       -2.1098     1.2345   -1.71   0.0912 .  

Coefficients:
                Estimate Std. Error z value Pr(>|z|)
treatment       1.2345     0.3456    3.57   0.0004

Standard Deviation vs Standard Error: The final match

Here we are with the big match between standard deviation (SD) and standard error (SE).

OK, let’s say it their names sound similar and both involve measuring variability, they serve fundamentally different purposes in statistical analysis. Understanding this difference is crucial for interpreting research findings correctly and avoiding common misinterpretations in biomedical literature.

Let’s start: what does the p-value depicts?

The p-value is a tool used to interpret statistical tests. It represents the probability that the observed results could have occurred under the assumption that the null hypothesis is true. In other words, it estimates the likelihood that what we are claiming is correct, within a small margin of error.

When comparing two treatments, like the classical treatment A and treatment B, the whole work starts with the assumption of the null hypothesis: that there is no difference between the two. The alternative hypothesis so that there is a difference and can be accepted if the null hypothesis is rejected.

In a nutshell statistical significance reflects how likely it is that the observed difference is real. There can never be absolute certainty, as a predefined margin of error must always be taken into account.

Conventionally, a threshold of p < 0.05 like a 5% margin of error is used to define statistical significance. This means there is less than a 5 in 100 chance that Treatment A would appear more effective than Treatment B purely by random chance.

• If p > 0.05: no statistically significant difference (null hypothesis is not rejected).
• If p < 0.05: statistically significant difference (null hypothesis is rejected).

The complexity of current diagnostic and therapeutic pathways requires highly specialized and tailored training, explains Paolo Marchetti, president of FMP and scientific director at IDI-IRCCS in Rome. Over the past fifty years, traditional teaching methods have seen very few changes, evolving from overhead projectors to electronic slides but without any real transformation in teaching practices. Plato represents a true leap forward, integrating advanced technology with a more interactive and engaging educational approach.

The research project initially focuses on oncology but aims to expand into other fields of medicine. Giuseppe Curigliano, president-elect of the European Society for Medical Oncology (ESMO), professor of medical oncology at the University of Milan, and director of the Division for New Drug Development for Innovative Therapies at the European Institute of Oncology (IEO) in Milan, believes that Plato has the potential to revolutionize medical education. Developed by a team of experts in education and artificial intelligence, it marks an important Italian contribution to the global landscape of advanced educational technologies. Plato may also be used to train members of Molecular Tumor Boards, multidisciplinary teams that are key players in the evolving model of precision oncology. The system is capable of analyzing genomic profiling data along with clinical information derived from precision oncology research within these boards.

The AI Learning study, promoted by FMP and hosted at Sapienza University of Rome, will start in September 2025. It will involve around 120 medical students in the final two years of their program.

Domenico Alvaro, Dean of the Faculty of Medicine and Dentistry at Sapienza University, emphasizes the institution’s strong commitment to experimenting with innovative teaching strategies to enhance the education of future healthcare professionals. In a time marked by the digital and technological transformation of health sciences, it is crucial to prepare highly skilled professionals in new technologies. The educational project coordinated by Professor Botticelli, along with several initiatives led by Professor Marchetti, perfectly fits into this vision.

In 2023, Novo Nordisk introduced NovoScribe, a new platform that automates much of the CSR writing process. NovoScribe uses retrieval-augmented generation combined with large language models like Claude 3, Titan, and a private version of ChatGPT. MongoDB Atlas plays a central role by managing the structured data that powers the system. By integrating clinical trial data with smart templates, the platform helps the AI models produce complete and accurate reports in just 10 minutes.

From such investments, Novo Nordisk has become the first pharmaceutical company to automate CSR generation at scale. Faster reporting not only helps the company accelerate regulatory submissions but also means patients can access new therapies more quickly. It is a major shift in how the pharmaceutical industry approaches reporting and data management.

Source: https://www.mongodb.com/solutions/customer-case-studies/novo-nordisk

The multifunctional biomedical robot can do imaging, sampling, and laser ablation.

Engineers at the Hong Kong University of Science and Technology have developed a new surgical robot, maybe the world’s smallest one. The 0.95-millimeter robot is 60% smaller than current endoscopic robots and is made of a hollow skeleton, optical fibers, functionalized skin, and a gel-like coating.The researchers say it achieves the “impossible trinity” of combining imaging, precise movement, and multiple functions. Their findings were published in Nature Communications.

The study showed the robot significantly expands the imaging area (25 times the normal view) and can detect obstacles up to 9.4 mm away (10 times the theoretical limit). It’s capable of sampling, drug delivery, and laser ablation, and navigated smoothly through in vitro bronchial models and ex-vivo pig lungs.

Professor Shen Yajing, who led the research team, explained that while small robots have potential for diagnosis and treatment, current models lack compactness, precise navigation, and diverse functions. Their work aims to solve these problems. Prof. Shen believes this robot could be a major step forward in clinical surgical robots, enabling early diagnosis and treatment in difficult-to-reach areas of the body, with broad biomedical applications. The team plans to conduct in vivo trials next.

The Asia-Pacific region is seeing increased investment in healthcare robotsics, especially for surgery and rehabilitation. For instance, Bangkok Hospital is investing $5 million in a robotic surgery center. The University of Hong Kong is using robot-assisted spine surgery. Chinese and South Korean wearable rehabilitation robots are now available in the US and Australia. In addition, even happening in Asia area, a Chinese team recently created a small, highly sensitive biosensor for continuous glucose monitoring.

The Bias in healthcare studies

When practicing evidence-based medicine, medical doctors and healthcare researchers often on data analysis to make informed decisions. However, one of the biggest challenges in research is bias—systematic errors that can distort findings and lead to incorrect conclusions.
Since then many choices in medical practice are made from fresh research papers it’s better to be well trained on get the idea of bias.
Here the main types of bias healthcare are explained with some examples, and offers strategies to reduce them. Bias in research can occur at various stages from study design to data collection and publication.

So, let’s find the bias when working in data analytics for evidence-Based Medicine.

Survival analysis

If you’re new to the world of medical research, you may have heard the term “survival analysis” thrown around and thought, “Sounds serious… is this about my career?” Don’t worry—it’s not (we hope!). Survival analysis is a statistical tool used to analyze time-to-event data, and it’s incredibly useful for understanding outcomes in medicine. Let’s break it down, step by step, with some examples.

In a nutshell, survival analysis answers questions like:

• How long does it take for an event to occur?

• What factors influence the timing of that event?

Here, the “event” could be a positive outcome (like recovery or a sneeze) or a less-than-ideal one (like disease progression or mortality). The key difference between survival analysis and other data analysis techniques is its ability to handle censored data. Censoring happens when you don’t observe the event during the study period—for example, a patient hasn’t died or relapsed by the time the study ends.

Fun Fact: Survival analysis isn’t just for medicine! It’s also used in engineering (how long before a machine breaks) and business (how long a customer stays loyal to a subscription). But in medicine, it’s a literal life-saver.

Common solution that Survival Analysis can gave us:

1. Cancer Treatment Trials

Imagine you’re testing a new chemotherapy drug. You want to know how long patients survive after starting treatment. Some patients may still be alive when the study ends, so you’ll have censored data. Survival analysis lets you account for this.

2. Organ Transplant Studies

You’re studying how long kidney transplants last before organ failure occurs. Here, time-to-event data (time until organ failure) is crucial to evaluating the success of different transplant techniques.

3. Time to Readmission

If you’re analyzing how long patients stay out of the hospital after discharge, survival analysis helps identify risk factors for readmission and assess interventions to reduce it.

Common Survival Models (Pick Your Poison, Statistically Speaking)

1. Kaplan-Meier Curves

The Kaplan-Meier method is like the “intro course” to survival analysis. It estimates survival probabilities over time and creates a curve showing the proportion of patients surviving at different time points.

Why Use It?

• Simple and visual.

• Great for comparing survival between groups (e.g., drug vs. placebo).

Example: You’re evaluating survival after heart surgery in two groups: those who received a new device vs. standard care. Kaplan-Meier gives you a clear picture of how survival curves differ.

2. Cox Proportional Hazards Model

The Cox model takes survival analysis to the next level. It examines the relationship between survival time and multiple predictors (like age, smoking status, or treatment type). It’s called “proportional hazards” because it assumes the effect of predictors remains constant over time.

Why Use It?

• Adjusts for confounding variables.

• Helps identify which factors significantly impact survival.

Example: In a cancer trial, the Cox model might reveal that tumor size and age significantly influence survival, while gender does not.

3. Parametric Models

Unlike the Cox model, parametric models assume a specific distribution (e.g., exponential, Weibull) for survival times. These are great when you’re interested in modeling the actual shape of the survival curve.

Why Use It?

• More informative when you can justify the assumptions.

• Useful for making predictions.

Example: You’re studying time until remission in a chronic disease. A parametric model lets you estimate not just median survival but also probabilities at specific time points.

4. Competing Risks Models

Sometimes, patients can experience more than one type of event, and these events compete. For example, in a study on cancer patients, death from heart disease is a competing risk for death from cancer.

Why Use It?

• Accounts for competing risks, avoiding biased results.

Example: You’re analyzing time to disease recurrence, but some patients die before recurrence occurs. Ignoring this competing risk would overestimate the recurrence risk.

Common Pitfalls when dealing with survival analysis (And How to Avoid Them)

1. Ignoring Censoring

Censoring is a cornerstone of survival analysis. Ignoring it is like ignoring patients who didn’t show up for their final follow-up—a big statistical no-no.

Pro Tip: Always check that your survival model properly accounts for censored data.

2. Overfitting

Using too many variables in your model can lead to overfitting, where the model fits the data too closely and performs poorly on new datasets.

Pro Tip: Be parsimonious with predictors. Use stepwise selection or shrinkage techniques to avoid this.

3. Assuming Proportional Hazards

The Cox model assumes that the hazard ratios are constant over time. If this assumption is violated, your results might be misleading.

Pro Tip: Check proportionality with diagnostic plots or use time-varying covariates if needed.

So, let’s Wrap It Up (Like a Survival Blanket). Why is Survival Analysis Useful?

Handles Censored Data: Not everyone experiences the event during the study period, but survival analysis doesn’t leave their data out of the equation. It’s like giving everyone a voice, even if they didn’t “finish the race”.

Flexible Models: Survival analysis offers a variety of models to suit your needs (more on that in a moment).

Clinical Insights: It helps you understand which variables (age, treatment type, comorbidities) affect time to an event, aiding personalized medicine.

Real-Life Applications: Whether you’re studying disease progression, treatment durability, or hospital outcomes, survival analysis has your back.

Survival analysis isn’t as intimidating as it sounds. It’s an essential tool for MDs diving into clinical research, helping you understand and predict time-to-event outcomes. Whether you’re drawing Kaplan-Meier curves or diving into the depths of the Cox model, remember: survival analysis is all about making the most of your data, even when life (or research) gets complicated.

So, the next time someone asks, “How long until the event occurs?” you can confidently say, “Let me run a survival analysis on that!” If they ask what that means, feel free to share this article — or just tell them it’s not about your career. (We hope…)

To go deeper on the topic:

• Kaplan EL, Meier P. “Nonparametric estimation from incomplete observations.” Journal of the American Statistical Association. 1958;53(282):457-481.

• Cox DR. “Regression models and life-tables.” Journal of the Royal Statistical Society: Series B (Methodological). 1972;34(2):187-220.

• Collett D. Modelling Survival Data in Medical Research. Chapman & Hall/CRC, 2015.

• Kleinbaum DG, Klein M. Survival Analysis: A Self-Learning Text. Springer, 2012.

• Goel MK, Khanna P, Kishore J. “Understanding survival analysis: Kaplan-Meier estimate.” International Journal of Ayurveda Research. 2010;1(4):274-278.

ChatGPT and his Artificial intelligence will make us live longer

OpenAI has recently announced an innovative partnership with Retro Biosciences, a biotech startup dedicated to advancing longevity research. This collaboration has resulted in the creation of GPT-4b Micro, a state-of-the-art AI model designed to expedite biological research focused on extending human lifespans by a decade or more. This initiative reflects a growing trend where artificial intelligence plays a pivotal role in biological and medical sciences, with OpenAI’s latest innovation poised to significantly impact longevity science.

Introducing GPT-4b Micro GPT-4b Micro is a specialized variant of OpenAI’s renowned GPT-4 model, tailored specifically for biological data analysis. Unlike the general-purpose GPT-4, this version has undergone extensive training on protein sequence datasets, enabling it to decipher protein interactions and suggest ways to manipulate them for research purposes.

The AI focuses primarily on a group of proteins known as the Yamanaka factors. These proteins have the remarkable ability to reprogram mature human skin cells into a more youthful, stem cell-like state. Scientists believe that by fine-tuning the behavior of these proteins, it may be possible to rejuvenate human cells, paving the way for longer lifespans and healthier aging.

A Novel Approach to Protein Engineering

GPT-4b Micro distinguishes itself from other AI models like Google’s AlphaFold through its unique approach to protein engineering. While AlphaFold excels at predicting protein structures, GPT-4b Micro takes a step further by generating novel protein variants for experimental testing. The model uses “few-shot” prompts—akin to how ChatGPT crafts text suggestions—to propose innovative combinations of proteins that might yield significant biological advancements.

By analyzing its extensive training data, GPT-4b Micro generates protein designs that researchers can validate experimentally. This methodology offers a transformative perspective, presenting scientists with creative experimental directions that may otherwise have gone unexplored. Essentially, GPT-4b Micro serves as a virtual architect for protein innovation.

Promising Early Results

Initial experiments with GPT-4b Micro have shown encouraging outcomes. Protein variants suggested by the AI have outperformed those designed by human researchers, with some achieving performance improvements of up to 50-fold for specific proteins. While these findings are promising, they require further validation through rigorous testing and peer-reviewed studies.

Should these results withstand scrutiny, they could represent a major leap forward in longevity science. Enhanced proteins capable of rejuvenating cells may significantly extend human life and healthspan, aligning with Retro Biosciences’ mission to leverage AI in the fight against aging.

Why It Matters

This development highlights the transformative potential of AI in accelerating discoveries in biology and medicine. OpenAI’s CEO, Sam Altman, has expressed optimism about AI’s ability to drive scientific progress, and GPT-4b Micro exemplifies this vision by contributing to advancements in longevity research.

Notably, Sam Altman has personally invested $180 million in Retro Biosciences, underscoring the growing interest and importance of this field. The collaboration between AI and biotechnology represents an exciting frontier in addressing humanity’s grand challenges, with profound implications for aging populations and global health.

Looking Ahead

Despite the promising early results, the journey to fully validate and refine GPT-4b Micro’s protein designs remains complex. However, the partnership between OpenAI and Retro Biosciences sets a precedent for integrating AI with biological research to tackle age-related challenges. By combining computational power with scientific inquiry, they are advancing efforts to extend human lifespans and improve the quality of life on an unprecedented scale.

As AI continues to revolutionize biological sciences, it is clear that these technologies will play an increasingly critical role in addressing some of humanity’s most pressing issues. While the ultimate impact of GPT-4b Micro on longevity research is yet to be determined, the collaboration between OpenAI and Retro Biosciences marks an important step toward a future where AI-driven solutions reshape our understanding of aging and health.

Is tech improving our healthcare journey?

The integration of digital technologies in healthcare is an evolutionary process, marked by both advancements and initial challenges. While the introduction of new technologies can lead to temporary setbacks and integration hurdles, the healthcare sector is currently experiencing a transformative period where data and technologies like artificial intelligence (AI) are converging to enhance processes, capabilities, and, most importantly, patient care. The following outlines some significant shifts occurring in healthcare delivery and medication management within complex supply chains and hospital systems.

From Basic Automation to Intelligent and Adaptive Assistance:

The 21st century has witnessed a surge in the adoption of automation in healthcare. However, early automation systems presented their own set of difficulties. While hospital pharmacies have realized certain efficiencies, these technologies have also introduced challenges, such as information overload.

Many communication systems, designed with good intentions, may have been implemented without a comprehensive understanding of hospital workflows. This has resulted in excessive and often unnecessary communication, contributing to staff burnout, particularly in pharmacy settings where personnel are bombarded with irrelevant or low-value notifications. Consequently, there’s a critical need for more user-centric and adaptive systems.

AI is paving the way for such systems, enabling greater customization based on individual user needs. Instead of being designed solely around a tool’s capabilities, features and notifications can be tailored to specific user requirements. This represents a significant advancement in hospital-based digital transformation, shifting the focus from simply having information available to optimizing how that information is accessed and utilized.

The importance of this customization becomes clear when considering the complex tasks healthcare professionals perform. For instance, an anesthesiologist in the operating room might be managing numerous concurrent tasks, including monitoring patient vitals, interacting with electronic health records (EHRs), and tracking medication administration. Currently, this information often originates from disparate sources. The goal is to consolidate and deliver this information to providers in a timely and relevant manner. The future envisions technology as an extension of the provider, potentially through innovations like virtual and augmented reality, providing real-time access to crucial information.

Enhancing Supply Chain Management through Predictive Analytics and Collaboration:

Effective pharmaceutical forecasting and management are essential for both operational efficiency and patient well-being. However, maintaining a balance between overstocking and stockouts can be challenging.

Drug shortages remain a persistent problem in healthcare. Conversely, excessive stockpiling can lead to waste through expiration or unnecessary inventory costs. Machine learning (ML) offers a promising solution for supply chain optimization. Centralizing data by integrating technologies across health systems creates a unified data repository. This repository provides real-time visibility into inventory, usage trends, expirations, recalls, and other critical data points.

By leveraging this comprehensive data, ML algorithms can generate analytical insights and predictive models to anticipate future needs and optimize ordering processes. This allows for more efficient distribution of pharmaceuticals within hospital systems, freeing pharmacists to focus on patient care.

Furthermore, cloud-based technologies facilitate inter-hospital collaboration. Moving away from siloed data centers enables greater data sharing and transparency. This is particularly beneficial for managing pharmaceutical supply chains, which can be unpredictable. Shared visibility can mitigate the risk of hoarding and ensure that critical therapies are available where they are needed most, especially during shortages or emergencies.

Prioritizing Ethical and Secure Transformation:

As hospitals embrace digital transformation, cybersecurity must be a paramount concern. Protecting sensitive patient information and ensuring the reliability and integrity of technology are crucial. Given the increasing sophistication of cyber threats, including AI-driven attacks, hospitals must be vigilant in vetting software and vendors, both new and existing. Legacy systems, often overlooked, also require thorough scrutiny.

Beyond technology assessment, establishing robust cybersecurity governance, risk, and compliance (GRC) programs and hiring qualified personnel are essential. All vendors should adhere to strict compliance standards.

When implemented thoughtfully and securely, digital transformation can enhance operational efficiency, reduce staff burden, and improve patient care. Ultimately, the greatest success will be measured by the positive impact on the patient experience.

The Bitcoin era

Different medical practices experience varying degrees of benefit from cryptocurrency adoption. Practices that primarily operate on a cash basis, such as outpatient clinics and those offering concierge or direct primary care (where patients pay a recurring fee for enhanced access), can enhance their modern image by accepting cryptocurrency. The growing field of medical tourism also stands to gain. For patients traveling from countries with unstable local currencies or unfavorable exchange rates, cryptocurrency offers a potentially stable and convenient alternative. For example, in El Salvador, Bitcoin is legal tender, and in Argentina, where inflation is high and dollar exchange rates fluctuate significantly, cryptocurrency can provide a way to navigate financial uncertainty.

While credit cards are popular for their convenience, their transaction fees (typically 3-5%) can significantly impact a practice’s revenue. Cryptocurrency payment gateways, however, often have much lower fees, sometimes as low as 0.5%. In the context of rising inflation and shrinking profit margins in healthcare, this difference can be crucial for maintaining profitability.

The potential of smart contracts, self-executing agreements coded on the blockchain, is a key aspect of this technology. These contracts automate transactions and reduce the need for intermediaries. In healthcare, they could streamline billing by releasing payments automatically upon service completion, such as after a telemedicine appointment, saving time and minimizing disputes.

Smart contracts can also improve regulatory compliance by embedding legal requirements directly into transactions, enhancing security and trust. While implementation requires technical expertise, the resulting transparency and efficiency can be valuable. For practices using cryptocurrency, smart contracts offer additional innovation and functionality.

Direct cryptocurrency transfers between patient and practice wallets, while technically possible and fee-free, are prone to errors and generally not recommended. Instead, practices should use cryptocurrency payment gateways. These platforms, similar to credit card processors, bridge the gap between patient wallets and the practice’s bank account, adding security layers. These gateways often mitigate price volatility by locking in exchange rates for a short period during transactions.

Upon receiving payment, the practice can choose to retain the cryptocurrency or convert it to traditional currency. Various gateways exist, each with different fees and supported cryptocurrencies. Some offer reduced or waived fees for premium memberships, which may be beneficial for practices with high cryptocurrency transaction volumes.

To begin accepting cryptocurrency, a practice can:

Create an account on a reputable and regulated cryptocurrency exchange (e.g., Coinbase, Binance). Integrate the chosen platform with their website or provide in-office QR codes for payments. Decide whether to hold cryptocurrency or convert it to traditional currency. Train staff to manage transactions and monitor payments using the platform’s interface. Many gateways, including Coinbase and PayPal, offer user-friendly app-based payment options.

When choosing which cryptocurrencies to accept, it’s important to be aware of the risks, including the prevalence of fraudulent or unstable currencies. Starting with Bitcoin (BTC), the most widely accepted cryptocurrency with the largest market capitalization, is advisable. Another option is USD Coin (USDT), a stablecoin pegged to the US dollar, which simplifies conversion to traditional currency.

Even major cryptocurrencies can fluctuate significantly in value. This can lead to patient dissatisfaction if the value changes substantially after a transaction. Therefore, obtaining informed consent from patients before accepting cryptocurrency payments and clearly explaining the potential for price volatility is crucial. It’s important to acknowledge that price fluctuations could benefit or disadvantage the patient.

While cryptocurrency offers a promising payment option for healthcare, it’s unlikely to completely replace traditional methods. Practices should maintain their existing payment infrastructure. Further research is needed to determine the prevalence of cryptocurrency use in healthcare transactions.

Cryptocurrency represents a potential evolution in healthcare finance. By adopting this technology, practices can modernize, potentially improve patient satisfaction, and streamline operations. Whether a practice is cash-based and seeking cost reductions or a medical tourism provider aiming to expand its reach, cryptocurrency offers potential benefits. A measured approach, continuous learning, and adaptability are key as this technology continues to develop.

AI and Medical journey

LLMs in medicine are nowadays super in hype. Or better, GenAI is super in hype. The last one is different from traditional AI in that it can generate content, such as summaries of text and answers to questions, rather than merely analyzing existing data. GenAI can definitively improve your work as healthcare professional since the goal is just to “removing the robot that is inside you to let you perform more human related task”. Hence the opportunity here of GenAI is just to manage exactly all the jobs that as healthcare professional you try to avoid every day to let you as much time as possible with patients. This since GenAI can empower your role, plus the patients to take a more active role in their healthcare. This since patients can use LLM-powered chatbots to interact with their EHR, prepare for appointments.

While challenges exist, the benefits of GenAI are too significant to ignore, and it may revolutionize care delivery by automating tasks, improving patient self-management, and enhancing job satisfaction.

By actively engaging in the development and implementation of GenAI, and adopting a balanced approach that prioritizes safety and responsibility, we can explore its potential to transform the way we deliver cardiovascular care and ultimately improve patient outcomes.

The article:

“It is a rather unexpected effect, because aspirin is mainly used as an anti-inflammatory drug,” says Marco Scarpa, a researcher at the University of Padova, and one of the authors of the study. As Scarpa notes, this study suggests that aspirin may be playing a slightly different role by stimulating the immune system’s surveillance response, which can then prevent or delay the progression of colorectal cancer.

Read full article (opens in a new page)