One of my goals as a science communicator with a focus on healthcare is to increase the general understanding of how to think about clinical interventions. Just take a look at the many health-related videos on TikTok or other social media platform, contradicting standard medical advice. Many are based on false information, but they also almost uniformly do not think about public health measures or medical decisions properly. This disconnect between how experts think of intervention and how the public may perceive them was massively on display during the recent COVID pandemic.

Here is the quick primer. Think of interventions based upon their risk vs benefit. You cannot only consider risk, and you have to consider the risk vs benefit of all options, including doing nothing. On the diagnostic end, it is better to consider the predictive value of information, not just whether it is a sign or symptom. In the realm of public health there are more considerations, not just the risk vs benefit at the individual and the population level. Because we have limited resources, and often there are competing valid options, we also like to look at the return on investment (ROI) of interventions. A recent study does that for the COVID-19 vaccines – how much public and private money was spent for their development, manufacture, and distribution and what were the health and economic benefits?

The bottom line is that the COVID-19 vaccination programs had a massive ROI, they had a positive cost-benefit and cost effectiveness (in addition to a positive benefit to risk ratio). But the details are interesting, so let’s dive in.

There are many ways to slice and dice such data, and many ways to estimate the cost benefits of preventing illness. So the authors provided as much information as possible, without just choosing one analysis. I am going to present the data from the bottom up – from the most conservative estimates to the most complete estimates. First, the estimated cost of all stages of the vaccine programs (including the first year of delivery) was $79.4 billion worldwide. That is a high price tag, but what did we get in return? The most straightforward type of cost benefit from a healthcare intervention is avoided future healthcare costs. The authors estimated this by using estimates of hospitalizations avoided and the average cost of hospitalization in each country. That amounted to $222 billion. So based on that maximally conservative estimate the vaccine programs were very cost-effective.  Continue Reading »

This question will not go away. I am getting asked this questions at conferences now. The media also seems to think this is a compelling question, and they are fueled in part by the tech industry itself, some of whom indulge this question. I will admit the answer has become more complex and nuanced than a simple “no”, but it is still mostly no.

Certainly the rise of neural networks, deep learning, transformer technology, and specifically large language models have ushered in the era of  – are we really sure current AI is not conscious? Our intuition tells us that they may be – LLMs are getting scary in how sophisticated they are and in their ability to simulate complex human conversations. Five years ago I was with the crowd who argued that LLMs are basically really good auto-complete apps, but this pithy sentiment no longer covers it. Even if they do not have a whisper of conscious awareness, they employ context-dependent logic. They are “intelligent” in their own way. And they are rapidly getting more impressive. So I get why people would be asking the question.

Everyone I know who works with AI apps has their example of when they were blown away by an interaction with an LLM. For me, I was trouble-shooting a web-based app I was using, and ChatGPT was very effective at helping with the trouble-shooting and producing a differential problem list. At one point it asked for a screen capture so it could have more information, and it gave me a cogent analysis. But here’s the part I found surprising – it “noticed” another window open in the background and inferred from this that I was also using this other app in the process, and used this to generate further potential sources of the problem.

Continue Reading »

The origin of life on Earth is one of the most difficult scientific questions to investigate. The emergence of living systems from non-living chemistry happened only once, as far as we can tell, 3.5-4.0 billion years ago. It’s a process that likely requires a lot of raw material (on a planetary scale) and a lot of time (up to 500 million years), and so is not something we can replicate in a lab. But life-origin researchers have made lots of progress none-the-less. One important question is where the prebiotic molecules that ultimately lead to life came from. For about a century one hypothesis has been that comets delivered much of this feedstock, in a steady rain over the course of the prebiotic era on Earth. However, a new study now casts doubt on that hypothesis.

Why was this ever a hypothesis in the first place? Why couldn’t all the chemicals necessary to develop into life have existed or been made right here on the Earth? The short answer is – they could, but there are unanswered questions and the cometary hypothesis would have been a convenient way to answer a lot of those questions. Prebiotic molecules must have included molecules with lots of carbon, hydrogen, nitrogen, and oxygen – all elements that would have been abundant on the early Earth. Researchers have demonstrated the plausible formation through chemical reactions of all the necessary components of life, either by creating abiotic synthesis in the lab or demonstrating viable chemical pathways. This includes amino acids, nucleobases, nucleotide precursors, lipids, sugars, and peptides. All the raw material was floating around in the “primordial soup” and could have spontaneously formed into all the building blocks of life. So what’s the problem?

Continue Reading »

Is the food category of “ultraprocessed” useful for consumers? To answer that I am going to take a huge detour, but I promise to return to this question.

Humans like heuristics. What this means is that we use a lot of cognitive short cuts to help us deal with a complex world in an efficient manner. Many of these heuristics are likely evolved, since they appear to be fairly universal. For example, we tend to look at the left most digit to get a quick estimate of the size of a number. That is exactly why marketers prices things as $19.99 rather than $20.00 – that really does work. There is also the availability heuristic – we estimate how common something is by how easily we can bring examples to mind. Since we don’t often have access to comprehensive statistics, this can serve as a useful metric.

The problem with heuristics, however, is that they are not strictly true or accurate. We sacrifice precision and accuracy for speed and convenience. This can be perfectly rational. If you can get 90% of the way to an answer with 10% of the effort, is it really worth making the other 90% of the effort to get that last 10% accuracy? In everyday life, probably not. The cost-benefit is not there, it will just slow you down for little gain.  If you are a scientist, however, then absolutely. That’s your job. But it seems to be also true that we evolved our common heuristics in a pre-industrial society, and now we are applying them to the extremely complex world we have erected for ourselves. So those heuristics break down much more often and much more catastrophically. We some some analytical critical thinking to adapt our thinking to the modern age.

Let’s apply the concept of heuristics to healthy eating. Most people are not nutritional experts, and yet in developed societies there is an increasingly complex array of food choices. Further, we evolved in a world where the primary nutritional concern was getting enough calories, and that’s pretty much it. Our bodies adapted to the food we had available to us, and we just had to keep from starving. Today people living in developed countries live with calorie abundance. Eating for optimal health, and an extended life expectancy, requires navigating often complex nutritional advice – glycemic index, trans-fats, vitamin levels, and macronutrient balance. Our evolved hunger, if you will, is not adapted to these choices. We have to use our frontal lobes.

Continue Reading »

Christina Applegate, an American actress, has been an outspoken advocate for patients with multiple sclerosis (MS) since her own diagnosis. She has been open about her “MS journey” in hopes that it will help others facing the same challenge. Recently, however, her journey took an unexpected turn – she apparently does not have MS. She reported on her recent podcast that her doctors informed her the diagnosis of MS made in 2021 is not correct. She lamented that now she is “back to square one”, without a specific diagnosis.

However, this does not invalidate her MS journey in my opinion. This is a very common aspect of that journey for many patients, up to 20% of which will turn out not to have MS even after an MS diagnosis. This is a great learning experience and there are several lessons worth pointing out. The first of which is that the fact that her diagnosis has changed does not mean that her neurologists did anything wrong or substandard. That seems to be everyone’s first reaction – a “misdiagnosis” means someone screwed up. This is not true, for very informative reasons.

A couple of background points before I push on. First, I recently retired from a 30 year career as a clinical neurologist. I was not an MS specialist, but I obvious know a lot about MS as a neurologist and was involved in the care of many MS patients. But MS diagnosis and treatment is extremely complex and should be done by an MS specialist, which we had in our clinic. Second, I have no inside knowledge of Applegate’s case, so I am basing my comments on publicly available information. Besides, most of my observations her are based on MS and clinical principles generally, not specifically her case. But it will put the news reporting into a useful perspective.

So how could Applegate have been diagnosed with MS when it turns out she does not have MS without this being a mistake? Diagnoses in medicine come in many different forms, and are based on different kinds of information. There are diseases, which generally result from a specific pathological process. There are also syndromes and disorders which may simply be a common clustering of clinical signs and symptoms. We use labels (diagnoses) for clinical entities that seem to exist in different patients, so that we can communicate about them, research them, and develop a body of knowledge about their natural history, diagnosis, and treatment. Some of these entities may be fairly iron-clad, based on objective definable pathology that can be documented unambiguously. At the other end of the spectrum are diagnoses that are simply symptoms that may have a typical demographic pattern and natural history, but without any known specific cause. There is also everything in between. There are also different types of data we can use – symptoms, signs on exam, blood tests, biopsy results, imaging studies, electrodiagnostic studies, response to treatment, and time course.

Continue Reading »

My wife and I had nothing scheduled yesterday, so we took the opportunity to vote early. On election day, everyone votes in their voting district. For early voting, the entire town votes in one location at the public library. When we arrived at the polling location we encountered long lines – longer than we have in past elections. Initially I thought this was due to unusually high early voter turnout, and that was likely partly the cause. But the main cause was completely unnecessary inefficiency in the voting process, which is what I really want to write about today.

Let me start by saying that voting is sacred to any democracy. Given that it is a critical right that should not be denied any citizen, it is the responsibility of the state to ensure that all voters have reasonable access to voting and the process should not be unduly burdensome. For this reason I think election day should be a national holiday. I also endorse early voting and voting by mail – voting should be as easy as possible within the confines of voting security. America has incredibly secure elections, with insignificant fraud. Since 2016 there have been only 513 proven cases of voter fraud out of 685 million votes.

But I am now talking about something which has nothing to do with election security, but rather with simple efficiency. For elections efficiency is mission critical, because we need millions of people to go through a secure system in a short amount of time. Anything that decreases efficiency even a tiny amount can not only inconvenience voters, it can potentially disenfranchise them. With that in mind, let me describe the process of voting I experienced yesterday.

There were three lines, separated alphabetically by street name. I get why this was necessary, because they want to keep a paper trail of who voted and this is organized by street. But this created unequal lines, and for much of my wait the middle line had no people in it, meaning that the workers at the corresponding table were idle while the others were backed up. They didn’t have a workflow to deal with this contingency. They also had inadequate signage – you could not read the signs at the front of each line from the back of each line. But that was something they were in the process of correcting while I was there.

Continue Reading »

It seems that the hottest topic of 2026 is the potential for artificial intelligence (AI) to bring about some terrible cataclysm, even including mass human death or extinction. This is like catnip to journalists and social media – it allows for sensational headlines and discussions based upon credible sources. What’s not to love? I agree AI is potentially a disruptive technology, and there are many AI-related issues we need to work out (data centers, intellectual property, displacing workers). But I have not seen a credible case from those arguing for near-term or inevitable AI doom. I also think that excessive doomism might short-circuit some of the potential benefits from AI.

To be fair, I also don’t buy the case for unfettered AI optimism. We have to be careful not to fall for the common human tendency to prefer simplistic and morally pure narratives. Whenever I discuss AI I almost always get pushback from those who cannot tolerate hearing anything positive, or others who do not want to hear anything negative, about AI. And when I try to balance my discussion, I get pushback for being self-contradictory. However, something can be both good and bad at the same time. It is not contradictory to point out that fact. And people need to wrap their heads around this fact – your morally pure narrative is not reality. The truth is somewhere between AI alarmism and AI utopianism.

A recent opinion piece makes a great analogy to nuclear power. Nuclear energy is a low-carbon source of reliable energy on a large scale. But nuclear weapons are potentially an existential threat to civilization, and the former can feed into the latter. Nuclear power can also be expensive, carries the risks of nuclear accidents, and creates spent nuclear fuel that we have to deal with. So you can focus on all the negatives to weave an anti-nuclear narrative, or all the positives to create an optimistic case for nuclear, but the reality is in between. Nuclear doomism arguably hampered the adoption of nuclear and exacerbated global warming. What we need is nuclear realism – it is a powerful technology with the potential for creating lots of low-carbon energy or for destruction, and it needs to be carefully managed and regulated. That’s pretty much what we have now, although we can do better.

Continue Reading »

It is an interesting time in human history for our understanding of life in the universe. We have the technology and basic knowledge to search for the signs of life on other planets, but we have not yet detected any. Confirmation of the high probability of life anywhere other than Earth would be an incredible scientific discovery, and it’s exciting to think it may happen in our lifetime. We now have multiple high-tech telescopes on and near Earth, so we are more capable than we have ever been before. But where to look?

We only have one example of life in the universe and so it makes sense to use that as a template. For this reason astronomers have been focusing on Earth-like planets around sun-like stars orbiting in the habitable zone.  The Earth is certainly habitable, as evidenced by the fact that it is teeming with life. But is it optimally habitable? Is looking for Earth twins the best place to look for alien life. Astrobiologists are asking this question, and developing a profile for so-called superhabitable worlds – those with optimal features for not just hosting life, but complex and potentially intelligent life.

I have mentioned in previous posts that orange stars (K-dwarfs) are likely the sweet spot for life. As stars get smaller and redder, they also live longer. We have a yellow star (by classification, even though the light it produces is white), which has a stable period of about 5 billion years. If life on Earth is typical, then this is just long enough to evolve intelligent life. Red dwarfs (M-dwarfs) live much longer, 100 billion to trillions of years. There are also a lot of them in the universe. But the light they produce is not ideal for photosynthesis, and they are very unstable, especially in their early lives. Also, the habitable zone is close enough to result in tidal locking. There is still the possibility of life on some of these worlds, but this is certainly not the ideal place to look.

Continue Reading »

The carbon nanofiber revolution has been slower than hoped. This is one of the drawbacks of being a technophile and following tech news – you hear about exciting technology years before it is realized, and many of the tantalizing breakthroughs never come to realization. So you have to constantly dampen your enthusiasm with caution and skepticism.

This has been the case with carbon nanofibers. The tech is exciting, based upon a two-dimensional material, carbon, aligned in a grid pattern (like chicken wire) just one molecule thick.  These can them be made into sheets, rolled into tubes, or woven into fibers.  We have had applications for this material for years, but they have been limited to carbon nanofiber “soot” – a collection of short strands that are added to other material to make it stronger or more conductive.

What we have not had is macroscopic organized carbon nanofibers – long strands of carbon cables or sheets of material. These can be made only in tiny amounts in the laboratory setting at great expense. This fits in with many exciting new potential technologies, they work well in the lab but the tech needs to be scaled up. This is often the deal-killer (and a detail often not mentioned in popular news about the technology). Tech news veterans, however, know that this is often the critical question – can it be scaled? Rather, how well can it be scaled, how does that match the demands of specific applications, and at what cost? Unless there are good answers to these questions you don’t have a disruptive technology, you have a lab curiosity.

Continue Reading »

The Nobel Prize is often awarded not to researchers who made a huge discovery about how the world works, but to those who have developed techniques that improve research itself. The downstream effects of such developments are likely to be dozens or hundreds, even thousands, of discoveries This is the case for the 2026 Nobel Prize in physiology or medicine – awarded to Karl Deisseroth, Peter Hegemann, and Georg Nagel for their work developing optogenetics. This is a powerful tool that allows us to create a functional map of the brain.

The story begins with Hegemann who in the 1990s was studying the free swimming algae, Chlamydomonas. The algae swims towards light, which anyone can reproduce by shining a light on a petri dish full of the green organisms. They have a small eyespot which can detect light, but Hegemann was fascinated with how quickly they were able to respond to a light signal. This lead him to hypothesize that the algae may be using a different method for turning light into a signal that can affect their behavior than vertebrate eyes. When he attached electrodes to Chlamydomonas and shined a light on them, he was able to detect an electrical signal in half a millisecond – incredibly fast. This is simply too fast to conduct any signal, so he thought it must be the same protein that detects the light and creates the resulting electrical pulse. Now all he had to do was identify that protein.

A Japanese group had already mapped the DNA of Chlamydomonas, and of the genes identified, two had properties that might fit the bill. But this was now a bit out of the field of expertise of Hegemann, so he contacted Nagel for help and sent him the code for the two candidate genes. Nagel injected the genes into separate batches of frog eggs, which began to express the proteins. He could then study how the proteins functioned, and was able to confirm Hegemann’s hypothesis. Both proteins, named channelrhodopsin-1 and channelrhodopsin-2, were ion channels that respond to light by opening up, allowing positive ions to flow through,  and creating an electrical potential. Channelrhodopsin-2 was particularly powerful, creating an electrical signal in 0.2 milliseconds. They were able to incorporate the gene for this protein into human and hamster kidney cells and show that they become light sensitive.

Continue Reading »