Melanoma: How Early Detection and Immunotherapy Are Saving Lives

Melanoma doesn’t always look like a scary mole. Sometimes it’s a tiny spot that doesn’t fade. Other times, it’s a bump that bleeds without injury. And if you miss it? The chances of survival drop fast. In 2025, the American Cancer Society reports that when melanoma is caught early-before it spreads-more than 99% of patients live at least five years. But if it’s found after spreading to other organs, that number plummets to just 32%. That’s not a statistic. That’s a life-or-death gap. And it’s why early detection and modern immunotherapy are no longer optional. They’re the only things standing between a routine check-up and a death sentence.

What Melanoma Really Looks Like (And What It Doesn’t)

Most people think melanoma means a dark, uneven mole. That’s true sometimes. But it’s not the whole story. Melanoma can be pink, red, or even skin-colored. It can look like a scar. Or a pimple that won’t go away. The ABCDE rule still holds: Asymmetry, Border irregularity, Color variation, Diameter larger than 6mm, and Evolving size or shape. But even that isn’t enough. Studies show primary care doctors miss up to 40% of melanomas during visual exams. That’s why tools are now stepping in.

One of the biggest breakthroughs isn’t a drug-it’s a camera. Devices like DermaSensor, approved by the FDA in early 2024, use near-infrared light to measure how skin scatters and absorbs energy. Normal skin and cancerous skin react differently. The device picks up those differences in seconds. In trials, it gave primary care providers an 85-95% sensitivity rate-meaning it caught almost all real melanomas. But here’s the catch: its specificity was only 26-40%. That means for every 100 healthy spots it flagged, 60-74 were false alarms. More biopsies. More anxiety. More cost.

AI Is Changing the Game-But It’s Not Perfect

Artificial intelligence is now reading skin images better than most humans. At Northeastern University, researchers built SegFusion, a system that first isolates the exact shape of a mole, then analyzes its texture and color patterns. It hits 99% accuracy. That’s not a guess. That’s from peer-reviewed data published in March 2025. It works by training on thousands of images, including ones from the ISIC 2020 dataset, where melanoma cases made up just 1.8% of the total. The system compensated by artificially boosting those rare cases so the AI wouldn’t ignore them.

Other models like DenseNet-201 and ResNet-50 have hit accuracy rates above 94% on standardized image sets. But real life isn’t a lab. Skin tones vary. Lighting changes. Moles get blurry. And most AI tools were trained mostly on light skin. A March 2025 JAMA Dermatology study found these systems perform 12-15% worse on darker skin tones. That’s not a small gap. It’s a dangerous blind spot.

That’s why systems like iToBoS, funded by the EU and led by Fraunhofer Institute, are trying something different. Instead of just looking at one mole, it scans your whole body in six minutes. It maps every spot. It flags anything unusual. And it doesn’t just say “cancer.” It shows the doctor *why*-using explainable AI. You see the heatmap. You see the edge detection. You see the comparison to known melanomas. That transparency builds trust. And trust matters when you’re deciding whether to cut into someone’s skin.

Wearable Patches and At-Home Screening

What if you didn’t have to wait for a doctor? What if your skin could tell you something was wrong-before you even noticed it?

At Wake Forest University, Dr. Mohammad J. Moghimi’s team developed a battery-free, wireless patch that sticks to suspicious moles. It doesn’t take pictures. It measures electrical resistance in the tissue. Cancer cells change how electricity moves through skin. The patch picks that up and sends the data to a small reader. In a pilot with 10 volunteers, the difference between healthy and cancerous tissue was statistically clear. The patch is comfortable. It’s cheap. And it could be used daily at home.

But it’s still early. Only 10 people. No large-scale trials yet. And it can’t tell if a spot is benign or malignant-just that something’s off. Still, for people with dozens of moles, or a family history of melanoma, it could be the early warning they need. The team is now testing conductive hydrogel electrodes to make the contact even better.

Immunotherapy: When Your Immune System Becomes the Weapon

For decades, melanoma that spread was a death sentence. Chemo didn’t work well. Radiation helped only temporarily. Then came immunotherapy.

In 2011, the FDA approved ipilimumab-the first drug to show it was possible to make the body’s own immune system attack melanoma. It wasn’t perfect. Side effects were brutal. But it worked. And it opened the door.

Today, the standard is combination therapy. Drugs like pembrolizumab and nivolumab block PD-1, a brake on immune cells. Drugs like relatlimab block LAG-3. And some patients get both PD-1 and CTLA-4 inhibitors together. These aren’t just treatments. They’re life extenders. In clinical trials, over half of patients with advanced melanoma saw their tumors shrink. A third had no signs of cancer after two years. That’s unheard of in the past.

Now, the next wave is coming. Regeneron’s fianlimab, paired with a PD-1 blocker, is showing even stronger results in Phase 3 trials. And IMA203 PRAME cell therapy, currently in the SUPRAME trial, is designed to train a patient’s own immune cells to hunt down melanoma cells with a specific marker. In early tests, 56% of patients had complete responses. That’s not a cure-but it’s close.

The Real Problem: Overdiagnosis and Inequality

More tools mean more detection. But more detection doesn’t always mean better outcomes. A 2025 paper in Taylor & Francis warned that we’re seeing a rise in overdiagnosis. That means finding slow-growing melanomas that would’ve never harmed anyone. We cut them out. We scar them. We stress them. And for what? No survival benefit. Just extra surgery.

And then there’s access. These new AI tools, full-body scanners, and smart patches cost money. Most are still in clinics in the U.S., Germany, and Japan. In rural Australia, or in low-income countries, they’re science fiction. Even in Sydney, not every GP has access to DermaSensor. And insurance doesn’t always cover it. The FDA has cleared 17 AI tools for melanoma since 2022. But getting them into clinics? That takes weeks of training, IT integration, and staff buy-in. One dermatology practice reported 20 hours of training per doctor just to use one system.

Meanwhile, the global melanoma diagnostics market hit $2.87 billion in 2024. Companies like DermTech are thriving. Google Health pulled its AI tool from the market in late 2024-not because it didn’t work, but because insurers wouldn’t pay for it. That’s the real barrier now. Not the tech. The money.

What You Can Do Right Now

You don’t need a scanner. You don’t need AI. You need to look.

• Check your skin monthly. Use a mirror. Take a photo of new or changing spots.
• Don’t ignore spots that bleed, itch, or don’t heal.
• Know your risk: fair skin, sunburns as a kid, family history, over 50-these raise your odds.
• If you’re high risk, ask your doctor about dermoscopy. It’s free or low-cost in most clinics.
• Wear sunscreen. Not just at the beach. Every day. UV damage adds up.

And if you’re worried about a spot? Don’t wait. Don’t Google it. See a dermatologist. Even if you think it’s nothing. Because melanoma doesn’t ask for permission. It grows silently. And when it spreads? It’s too late for most.

The Future Is Here-But It’s Not Fair

By 2030, AI-assisted detection could become standard. Immunotherapy will keep improving. We might even see blood tests that catch melanoma before it shows on the skin. But progress won’t help everyone equally. The tools exist. The science is solid. What’s missing is equity.

For now, the best defense is still you. Know your body. Speak up. Get checked. And if your doctor dismisses you? Find another one. Melanoma doesn’t care if you’re rich or poor. But your survival? That depends on how fast you act.