Kumbh Mela: AI reveals pattern in stampedes

Kumbh Mela: AI reveals pattern in stampedes


On January 29, the world’s largest religious gathering, Maha Kumbh in Prayagraj, was struck by a horrifyingly familiar tragedy — a stampede that claimed 48 lives.

This despite decades of experience in organising similar congregrations, improved technology, and advanced crowd-monitoring systems. A stampede during the Kumbh in 1954 claimed no less than 700 lives.

A study titled ‘At the Mahakumbh, Faith Met Tragedy: Computational analysis of stampede patterns using machine learning and NLP’, by Abhinav Pratap of Amity University, examines the recurring failures in crowd management at India’s grandest pilgrimage. Applying machine learning, historical data analysis, and natural language processing (NLP) to seven decades of administrative records, the research uncovers a disturbing reality: stampedes at the Kumbh Mela are not accidents but predictable failures, resulting from infrastructural limitations, governance inertia, and a tendency to normalise disaster. 

Integrating computational modelling and sociological theories, the researchers designed a three-phase analysis to derive insights from past tragedies, including patterns of administrative failure and recurring risk factors. They then placed the findings within the framework of the Emergent Norm Theory (which explains how collective behaviour in large crowds overrides individual rationality) and Institutional Amnesia Theory (which describes how organisations, institutions, or even societies gradually lose their collective memory of past events due to staff turnover and lack of proper documentation). 

A recurrent crisis

The analysis of stampede incidents between 1954 and 2025 throws up a striking pattern — critical crowd density thresholds (equal to or more than seven persons per sq m) consistently lead to deadly outcomes. Once this limit is breached, individual control is lost, and panic spreads through the crowd like a chain reaction. 

VIP movement

One of the most alarming findings is that 92 per cent of past stampedes occurred near infrastructural choke points — narrow pathways, riverbank access routes, or restricted entry points that become dangerously overcrowded. 

The study also highlights the role of restrictions due to VIP movement, which redirect police forces and create dangerous bottlenecks. Even in 2025, drone footage revealed barricaded exits left unmanned during a ministerial visit, exacerbating the deadly crush. 

The NLP-driven analysis of official stampede inquiry reports from 1954 to 2025 repeatedly show administrative narratives that deflect blame from authorities:

• 1954: “Unforeseen surge” 

• 1986: “Crowd became unruly” 

• 2003: “Poor coordination” 

• 2013: “Railway station mismanagement” 

• 2025: “Barricade collapse”

Instead of acknowledging systemic failures, these reports frame stampedes as random and unavoidable incidents.

Improved governance reduces fatalities, but its impact remains limited due to systemic failures. 

In this case, there were three failures: Delayed emergency response, as reflected by the inquiry reports from 1954, 1986, 2003, and 2025; VIP route prioritisation disrupts normal crowd flow, creating bottlenecks that intensify stampedes; and AI-based risk predictions were ignored, leading to reactionary, rather than preventive measures. 

The 2025 stampede occurred at a barricade breach where drone footage showed unmanned exits — a repeat of the failure from 1954.





Source link

Tackling stillbirth crisis with pooled datasets

Tackling stillbirth crisis with pooled datasets


The Indian Council of Medical Research (ICMR) has launched a groundbreaking initiative to address the country’s persistent stillbirth crisis. The study collates medical records of 2.29 lakh pregnancies across nine states. According to the World Health Organisation (WHO), a stillbirth is when a baby is born without signs of life after 28 weeks of pregnancy. Currently, India records 14 stillbirths per 1,000 births and aims to reduce this to fewer than 10 by 2030.

The ICMR-Stillbirth Pooled India Cohort (ICMR-SPIC) represents an unprecedented collaboration in Indian medical research. Drawing from 10 research groups across 17 locations, the study captures the diversity of India’s population. This broad scope allows researchers to identify patterns and risk factors that smaller studies may miss. “If a single dataset cannot cover the diversity needed in the data… it is better to have pooled datasets,” explains Assistant Professor Palash Ghosh from IITGuwahati’s Department of Mathematics. 

Pooled methodology is particularly valuable in studying complex health issues like stillbirth, where multiple factors — including location, genetics, and economic status — influence outcomes. “With diverse populations, the conclusions made from the data become truly representative,” says Ghosh. 

The project, despite its scale, faces several challenges. These include variations in data collection methods, incomplete information about stillbirth timing, and inconsistent records of medical care during delivery. 

To ensure reliability, the ICMR-SPIC consortium reports that rigorous data standardisation procedures were implemented. Researchers reviewed and harmonised information across studies to maintain consistency. 

The ICMR-SPIC initiative looks to translate research into practical tools for medical professionals. The study will help develop predictive models for identifying high-risk pregnancies, creating guidelines for targeted interventions, and establishing frameworks for clinical decision-making.

(Yasaswini Sampathkumar is a writer based in Guwahati)





Source link

Amino acids that can beat hypertension

Amino acids that can beat hypertension


High blood pressure, or hypertension, affects about 320 million Indians. Untreated, the condition could lead to cardiovascular complications, trigger kidney issues and cause eye trouble. 

Recent research by Prof Nitish Mahapatra and his team at IIT-Madras has shown how a peptide and its variant have vastly different effects in the natural control of hypertension. 

A peptide is a short chain of amino acids that serves as a building block for proteins. Catestatin is a peptide made up of 21 amino acids. The study focused on how the normal form of catestatin (called wild type, or CST-WT) and a common variant (Gly364Ser, CST-Serine) affect cardiovascular function. 

But what made the researchers focus on these peptides in particular?

Says Prof Mahapatra, “We narrowed down on them because we had published a paper in 2016, in a journal called Hypertension, where we showed that people harbouring the variant peptide were more prone to developing hypertension.” 

The study had analysed about 4,000 people across India and found that 15 per cent carried the serine variant. Their risk of developing hypertension was about twofold greater that those with the wild-type. 

Their latest paper is the culmination of the 2016 study. “In the animal model, we saw that what we had hypothesised was indeed correct,” he says. “We found the wild-type form of the peptide to be antihypertensive in our laboratory.” 

Heart function test

Hypertension, or high blood pressure, is a condition where the force of blood against the walls of blood vessels remains high over time. 

In the IIT-Madras study, rats were fed a high-salt diet for a few months. The animals, which had become hypertensive, were injected with the normal or wild-type peptide. It was observed that their blood pressure dropped to normal within two hours, and the effect usually lasts 24 hours or so. On the other hand, when the serine variant was injected the blood pressure remained high.

The study also measured heart rate and observed the electrical activity of the heart using an electrocardiogram. High heart rates are known to be associated with heart diseases. 

The wild-type form of the peptide was able to reduce the heart rate in hypertensive rats, whereas the serine variant showed a smaller effect. In addition, improvement in heart muscle contractions, which reflect how strongly the heart pumps blood, was greater with the wild-type peptide than with the serine variant. 

Measurements of pressure in the heart both during contraction and relaxation showed that the CST-WT peptide helped restore normal heart function better than the variant. 

The heart tissue from hypertensive rats also showed evidence of inflammation and cell damage. Rats treated with the wild-type peptide had less tissue damage and fewer signs of inflammation than those treated with the serine variant.

One step ahead

So, are the researchers working with pharmaceutical firms to formulate a drug that addresses high blood pressure? 

“Right now, we are doing something different,” Mahapatra says.

After its latest research was published, the team is modifying amino acids of the wild-type peptide to make it even more potent and stable. The study focused on naturally occuring peptides — both wild-type and serine, Mahapatra explains. 

The effort now is to modify them to help keep blood pressure under control for a longer period of time.

Eventually, that part of the team’s work would go for human clinical trials in collaboration with pharmaceutical companies.





Source link

AI implementation in healthcare requires tailored approaches: ISB Study

AI implementation in healthcare requires tailored approaches: ISB Study


AI implementation in healthcare requires tailored approaches considering factors such as regional infrastructure differences and training programmes. 

As per the findings of a study by researchers at the Max Institute of Healthcare Management at the Indian School of Business on ‘Understanding Providers’ Attitude Toward AI in India’s Informal Health Care Sector,’ 93.7 per cent of providers believed AI could improve TB diagnosis accuracy, while only 69.4 per cent were willing to adopt the technology. 

Sumeet Kumar, Assistant Professor, Information Systems, ISB and lead author of the study said: “The gap between belief in AI’s potential and willingness to adopt it suggests that technological superiority alone may not guarantee successful implementation.”

Regional differences and existing healthcare infrastructure play crucial roles in technology adoption, he added. 

The key findings of the research include higher adoption readiness in Gujarat (73.4 per cent) compared to Jharkhand (58.4 per cent), reflecting the impact of regional healthcare infrastructure development. Providers who were more confident in diagnosing TB showed greater willingness to adopt AI. Also, providers’ trust in local radiologists influences AI adoption differently across regions.

The research suggests that successful AI implementation in healthcare requires tailored approaches considering regional infrastructure differences, additional support and training programmes for healthcare providers, focus on providers with limited access to diagnostic infrastructure and consideration of cost implications.

The study surveyed 406 Ayurveda, Yoga and Naturopathy, Unani, Siddha and Homeopathy and informal healthcare providers (collectively called AIPs) across Gujarat and Jharkhand. 

TB remains a global health crisis, claiming 1.5 million lives in 2020 alone, with India bearing a substantial burden. Early and accurate diagnosis is key to TB treatment, but accurate diagnostic tools like molecular diagnostic tests are expensive, difficult to access and challenging to maintain, ISB release said. 





Source link

How to read burnt scrolls

How to read burnt scrolls


On August 24, 0079, Mount Vesuvius blew its top. The citizens of Pompeii and Herculaneum were caught off guard by the hellfire. The city perished under hot ash. But the killer ash also ended up entombing the city and preserving it near intact.

Nearly 1,700 years later (in 1738), Herculaneum showed up when workers were digging for a well. King Charles III of Spain, who was also the king of Naples, lost no time in ordering excavations. Ten years later, when they unearthed Pompeii, they found the city remarkably well preserved. 

Pompeii has since become the stuff of fantasy and folklore, inspiring books and movies; but, importantly, it has served as a trove of knowledge, revealing much about the past. But it has also held several mysteries — one of which are a set of rolled and charred papyrus. 

How on earth can one decipher what a charred papyrus says? With artificial intelligence. 

So you now have an open ‘Vesuvius Challenge’, offering money to those who can reveal the blackened words. Some have even won good money for uncovering a few words here and there. 

But how will AI “unroll” a rolled papyrus? We put the question to AI itself (ChatGPT) and it replied as follows: “The process typically starts with a high-resolution scan (like a CT scan) that produces a detailed 3D digital image of the scroll’s interior structure. Then, AI algorithms — often using techniques from computer vision and deep learning — analyse these images to identify the different layers of the scroll and the areas that contain text. Once the layers are segmented, mathematical models and geometric transformations are applied to “flatten” these curved surfaces into a 2D plane. This virtual unrolling preserves the text and structure without physically handling the fragile artifact.”





Source link

Fake cells get real work done

Fake cells get real work done


Scientists have been working for decades to create synthetic cells that mimic key functions of biological cells. The challenge has been to make the artificial cells dynamic, responsive, and capable of interacting with their environment in a controlled way. A new study published in Nature Materials, ‘Morphology remodelling and membrane channel formation in synthetic cells via reconfigurable DNA nanorafts’ (authored by Sisi Fan, Shuo Wang, Longjiang Ding, and others), presents a breakthrough involving the use of DNA nanorafts — tiny, flat assemblies of DNA strands that change shape in response to chemical signals.

By using DNA as a programmable building material, researchers have found a way to engineer membranes that can change shape, open and close transport pathways, and respond to specific molecular signals.

Molecular marvel

DNA’s ability to form predictable and programmable interactions makes it an ideal material for building nanoscale devices, such as the nanorafts used in this study. 

The nanorafts were integrated with giant unilamellar vesicles (GUVs), which serve as simplified models of biological cell membranes. By attaching DNA nanorafts to the surface of these synthetic cells, scientists were able to induce and control changes in their membrane structure. Membrane shape is critical to many cellular processes, including transport, engulfing nutrients, sending signals, moving through tissues, and division.

The innovation here is that the DNA nanorafts are not passive structures. They are designed to switch between different shapes and, in doing so, they actively influence the behaviour of the membranes they are attached to, marking a major advance in synthetic biology. It opens up new possibilities for engineering artificial life-like systems.

Cell shaping

Initially the nanorafts are in a square-like conformation; but when exposed to specific DNA strands called ‘unlocking strands’, they elongate into a rectangule, which applies force on the membrane and deforms it in a controlled way.

Equally important is the ability to reverse these shape changes using a different set of ‘locking strands’.

Gatekeeping

Beyond reshaping membranes, the DNA nanorafts enable another crucial function: forming and sealing synthetic channels within the membrane. Transporting molecules across cell membranes is one of the fundamental functions of life, allowing cells to absorb nutrients, remove waste, and communicate with their surroundings. In biological cells, this process is carried out by protein-based channels and transporters. However, engineering such complex protein structures has always been a challenge in synthetic biology.

The study offers DNA-based structures as an alternative. When the nanorafts elongate, they interact with the membrane in a way that creates openings or pores. These synthetic channels allow in molecules as large as 70 kilodaltons (kDa), which is not possible in most natural protein channels.

Importantly, these pores are not permanent. Once the membrane regains its original shape, the pores close and seal off the transport pathway. This control over membrane permeability is a major step forward in creating artificial cells that can selectively regulate molecular transport.

Applications

The ability to reshape membranes and create selective transport channels has significant implications for drug delivery, bio-sensing, and artificial cell research. Imagine a synthetic cell that can absorb a drug molecule in one environment, travel through the body, and release it at a specific site, such as a tumour. Programmed DNA nanorafts promise highly precise drug delivery with minimal side effects.

Synthetic cells can also serve as biosensors to detect disease markers, toxins, or bacterial infections by responding to specific molecules and changing their membrane structure accordingly. They could provide real-time medical and environmental diagnostics.

In fundamental research, DNA nanorafts can help us understand how biological membranes function and gain new insights into the fundamental principles of cell biology.

While natural cells have evolved complex mechanisms for regulating their shape and transport processes, the new findings suggest that we can design artificial systems with similar and, potentially, enhanced capabilities.





Source link

YouTube
Instagram
WhatsApp