At CAIAL—Center for Advanced Imaging & Longevity, we detect health issues early with high‑res imaging and screening, then guide your care for better health.
Do not take daily tadalafil or sildenafil routinely for longevity, “vascular improvement,” cancer prevention, or other unrelated uses—especially if you are a healthy man without symptoms.
These medications have established indications. Tadalafil is used for erectile dysfunction, symptoms of an enlarged prostate, and pulmonary arterial hypertension at indication-specific doses. Sildenafil is used for erectile dysfunction and pulmonary arterial hypertension.
There is no reliable randomized, placebo-controlled trial evidence that taking these drugs routinely prevents heart attacks, strokes, or cancer—or extends life. The new cancer research involved cells, mice, and observational health records, not a randomized human treatment trial.
These drugs can also have serious interactions with nitrates, riociguat, alpha-blockers, blood-pressure medications, certain antifungal and antiviral drugs, and heavy alcohol use.
The message is simple: do not use these drugs for vascular health, cancer prevention, or longevity until real clinical trial data demonstrate that the benefits outweigh the risks.
Do not base medical treatment on social-media theories, personal enthusiasm, or observational associations presented as proof. A promising theory is not the same as proven medicine.
Imaging guided treatment of subclinical atherosclerosis—We’ve come a long way! - Journal of Clinical Lipidology
@EugeniaGianos@ACCinTouch@nationallipid https://t.co/h6aDFypjGT
Yes, we are past the point of no return.
When debt service payments squeeze out spending, it is like plaque in the circulatory system squeezing out the flow of blood. It is the same kind of thing, and it can be measured. We are seeing that happen right now.
There is also a supply and demand issue. A budget deficit means that debt has to be sold. We can see this happening in the bond market—bonds have been a bad investment, and there is pressure on interest rates and borrowing.
This dynamic is happening, but people are treating it like it hasn't happened before. They don't understand that, just like plaque in the arteries, the debt builds up and they have that exposure.
My gmail got hacked. I am an idiot. Glass half-empty, I am an idiot and have a mess to clean up. Glass half-full, I have heard from dozens of people I had not seen or spoken with in forever
@cvmg and @niCVRI began implementing coronary CTA before invasive angiography in 2006 and deeply appreciate your @ColletCarlos advocacy of this approach. Invasive coronary angiography should generally be reserved for patients with unstable angina, acute myocardial infarction, or cases in which CTA, physiologic assessment, and the patient’s symptoms do not provide enough information to make a confident clinical decision. It took nearly two decades for this principle to become firmly established in the evidence base: Integrated Coronary Computed Tomographic Angiography in an Office-Based Cardiology Practice: https://t.co/s8poAFUQc2 @niCVRI #plaquedirectedtherapy @BudoffMd
I was so honored to spend time with key mentors during my 18 years at @MassGenBrigham, @broadinstitute, and @harvardmed, as I embark on a new chapter at @Amgen. From them, in addition to learning how to be a physician and a scientist, I learned, and continue to strive, to motivate and collaborate with humility and curiosity. This has been the most wonderful adventure of continual growth and exploration, surrounded by the most wonderful people.
Grok defines the best open AI models: The world has changed in a week. Medicine is impacted! Your future health care got better and less expensive!
As of July 2026, there isn’t a single “best” open-source (or open-weight) AI model—it depends on use case, hardware, license, and benchmarks—but top contenders include GLM-5.2, Kimi K3/K2 series (Moonshot AI), DeepSeek V4, and Qwen3/Qwen3.5 variants. These Chinese-led open-weight models have narrowed or closed the gap with closed models like GPT-5 or Claude in many areas, especially reasoning, coding, and agentic tasks.
Top Recommendations
• Best overall/versatile: GLM-5.2 (https://t.co/mTIJFkyeXD)
Frequently ranked #1 for long-horizon coding, agentic workflows, repository work, reasoning (e.g., high GPQA scores), and general capabilities. Strong on long context (up to 1M tokens in some variants). MIT license makes it very usable.
• Best for coding/agents: Kimi K3/Kimi K2.7 Code or DeepSeek V4-Pro
Kimi leads on several leaderboards (e.g., Humanity’s Last Exam ~56%) and coding benchmarks (high SWE-Bench). DeepSeek excels in math, reasoning, and software engineering. Great for developers.
• Strong all-rounder/cost-effective: Qwen3/Qwen3.5 (Alibaba)
Excellent balance of performance, efficiency (MoE architectures with fewer active params), Apache 2.0 license, and hardware friendliness. Strong in coding, instruction following, and multilingual tasks. Good for local runs.
Other notables:
• MiniMax M3/M2: Strong in agentic/coding workflows.
• Llama 4 (Meta): Massive context windows (up to 10M tokens in Scout variant) for long-document/RAG tasks; solid but often not the absolute leader in raw benchmarks.
• Smaller/efficient options like Gemma 4 for laptops/edge devices.
Quick Comparison (Approximate from 2026 Leaderboards)
@RonKarlsbergMD Early screening and monitoring of aortic aneurysms can protect decades of longevity—rupture is often fatal, but proactive care turns a silent threat into a manageable condition. @RobertKennedyJr