This glioblastoma treatment sounds both unreal and very satisfying

Brain cancer is hard to treat by just cutting it out, and it can become resistant to drugs. So what’s left? I dunno … how about shredding it from the inside?

A group of Canadian researchers came up with this idea. They took carbon nanotubes and filled them with iron oxide (please don’t call it “rust”). They coated the outside with an antibody that would bind to glioblastoma cells. Then they injected it, and the cancer cells gobbled up the tubes.

Remember the iron filling?

[B]y activating a magnetic field near the cancer cells, the tubes were made to spin, wreaking havoc to the internal structure of the cells — particularly to their mitochondria, which fundamentally provides cellular energy. In effect, the tubes acted like thousands of mini scalpels that sliced up the cancer cells from the inside.

This may be one of the most satisfying ways to kill cancer. As Son of Buzz put it, “It’s like a pipe bomb in the mailbox.” It worked on mice (it always does, doesn’t it?), but needs more fine-tuning before it can be tested on humans.

Poor diet, too much diabetes

There are — per a giant “research model of dietary intake” out of Tufts University — something like 14 million cases of type 2 diabetes across the world were caused by poor diet. And that was in 2018.

But “poor diet” is a bit general. When they drilled down, they found that it’s specific eating habits that are the biggest problem.

Insufficient intake of whole grains, excesses of refined rice and wheat, and the overconsumption of processed meat [had the biggest impact].

In contrast…

Factors such as drinking too much fruit juice and not eating enough non-starchy vegetables, nuts, or seeds, had less of an impact on new cases of the disease.

So switch from white rice to brown, whole wheat is better than white bread, and go easy on meat that’s been salted, cured, fermented, smoked … you know, more than just meat. And then convince your patients to do the same.

Why don’t people take statins?

Sure, you can ask patients why they don’t take their meds, but — to quote House, MD — patients lie. To healthcare pros, at least. But where might they tell the truth? Social media.

That was the premise of Stanford U researchers. So they asked an AI to analyze statin-related posts and comments on Reddit, the big ol’ public collection of message boards. In a shock to absolutely no one, “They found that discussions about statins on the platform are mostly negative and rife with misinformation.”

As one Standforian put it, “[W]e saw a lot of misunderstandings about side effects and unproven alternative treatments to lower cholesterol.”

In general, people were iffy about statins because they thought pharma companies had manipulated clinical trials of the drugs, or that the danger of LDL cholesterol was overblown because they felt better after losing weight using a high-fat, low-carb keto diet.

The research paints a troubling picture of patient sentiments toward potentially life-saving drugs and evidence-based medicine generally.

Small study suggests phages work as a last resort

When antibiotics fail due to resistant bacteria, there’s not a lot left in the toolbox. Bacteriophage therapy is one possibility, though — finding the virus (i.e,. the phage) that matches the problem bacteria. If you can do that, you might have another treatment option.

This isn’t news, and it’s been done. (The US Navy in fact has one of the largest phage libraries in the world.) But there’s not much data on how effective it is.

Now Israeli doctors have at least a small study. Since 2018, they received 159 requests for last-ditch-effort phage therapy — compassionate-use treatments. They were able to find and use a matching phage in only 18 cases, though. We told you it was a small study. That said…

Of the 18 patients who received intravenous phage therapy, 14 (78%) achieved clinical remission, and 4 (22%) were classified as treatment failure. No major side effects were reported.

There is no official protocol for phage treatment, and matching phages to bacteria is hit-or-miss, so for the moment this kind of data ‘drip’ is the best we can hope for.

60-second schizophrenia test

What if you could diagnose schizophrenia in less than a minute simply by having a patient stare at a fixed object? That’s what Chinese scientists say they can do, using high-speed camera and a bit of artificial intelligence.

The gist: They measure saccades — tiny eye movements — while a patient stares at a target for about 60 seconds.

They found that patients with schizophrenia exhibited significantly more vertical saccades and a greater vertical deviation of horizontal saccades.

It got even more detailed: The longer a horizontal saccade lasted, the lower the patient’s cognitive performance, and the more it deviated vertically the more severe were the symptoms.

How accurate were these measurements? Fed the saccade data, the AI’s diagnosis was correct 85% of the time. Whoa.

Short Takes

Electric bedside manner

When patients have questions about their health conditions, responses from an AI were rated (by healthcare pros) to be better than those of a human physician 79% of the time. Not only did the AI respond “with nuanced and accurate information that often addressed more aspects of the patient’s questions than physician responses,” but…

ChatGPT responses were rated significantly higher in quality than physician responses: good or very good quality responses were 3.6 times higher for ChatGPT than physicians. The responses were also more empathic: empathetic or very empathetic responses were 9.8 times higher for ChatGPT than for physicians.

Sleep, dementia, and dogs

Dogs who have dementia suffer from the same kind of sleep disturbances as humans. And when they do sleep, it’s not really sleeping: “[W]e saw their brain activity during sleep was more akin to wakefulness.” (As we mentioned in a Short Take a week or so ago, there’s a vicious cycle for sleep disturbance and symptoms of dementia.)

Side note: Play with AI yourself

You can create a free account and play with ChatGPT, the OG of artificial intelligence, at chat.openai.com.

Want it to write something for you? Give it detail: “Compose a letter to XYZ Corporation asking for more data about their ZYX medication, specifically how it affects older patients with cognitive decline” or something similar. The result won’t be perfect, but it will probably give you a good start.

You can also ask it to analyze text. Paste in a paper and write, “Summarize this text in 150 words or fewer.” Just for fun, we pasted in a couple of Buzz items and asked it to “Describe the writing style and quality of this blog post.”

The writing style in this blog post can be described as informative, concise, and conversational. The author presents various medical findings and breakthroughs in a straightforward manner, using simple language and short paragraphs to keep the reader engaged. The author also injects some humor and personality into the post, as seen in the parenthetical comments and witty titles for each section. Overall, the style is easy to read and understand, while still conveying important and interesting information about medical research.