The appointment screen said April was being brought in to follow up on the weekend’s A&E visits. I clicked into her notes: an 18-month-old girl, nothing remarkable except two trips to the emergency department in the space of 48 hours.
The Saturday discharge letter said April had been running a high fever for several days. The doctor had found an inflamed eardrum and prescribed amoxicillin. Back she came on Sunday, covered in a rash. The second casualty officer had changed her to a different antibiotic, suspecting an allergic reaction.
I called the family in – both mum and dad were there. As soon as she’d taken a seat, mum put the two medicine bottles on my desk: “That was the first one, this is the second. But look.” She lifted her daughter’s vest up, revealing an impressive patchwork of red blotches all over her skin. “It’s even worse.”
I got them to talk me through the story: four days of unrelenting fever, never dropping below 39°C, until eventually they’d lost their nerve and taken her to the hospital. I confirmed that she’d had no other symptoms, just the temperature. A scattering of spots had appeared the day they went to A&E, but they’d dramatically worsened after starting the first medicine and become more florid with the second. Dad gesticulated at the bottles. “We don’t know what to do.”
While the parents were talking, I kept my attention on April. She was sitting happily on her mother’s lap, looking at her new surroundings with interest. Even though I was sure of the diagnosis by then, I checked her over. Neither ear had any sign of infection. Eardrums will redden with high fevers – just as skin flushes – and doctors anxious to find an explanation can seize on the idea of an ear infection as the cause.
I steeled myself: I was about to give them yet another opinion, and I feared it might be an uphill task to persuade them that mine was the one to be believed. “I know it gets confusing when you see lots of doctors and get told different things,” I said, “but sometimes a diagnosis only becomes clear over time.” I explained that April had never had an ear infection; both antibiotics had been unnecessary but neither of them had caused the rash. Everything was explained by a viral infection that was going to get better on its own.
As I was talking, I wrote the diagnosis on a Post-It – people can go home and read up on it online, and I knew when they did that, everything would make sense to them. “It’s got a bit of a weird name,” I said, handing the note across. “Roseola infantum.”
Dad sat up straight in his chair. “That’s what AI said!” He must have seen the surprise on my face. “You know how you can upload pictures? We put one of her rash in and that’s what it said.”
Roseola is one of my favourite diagnoses. Historically it was known as sixth disease, being one of six infections of childhood – others include measles and scarlet fever – that can be distinguished by virtue of their characteristic rashes. We see a lot of roseola in general practice (A&E doctors see it less so) and becoming adept at spotting which toddlers with the roiling temperatures can be safely left alone saves much parental anxiety and unnecessary intervention.
AI is brilliant at pattern recognition, and it impressed me that it had interpreted April’s skin eruption so accurately. But it hadn’t been enough for her parents. The mixed messages and antibiotic prescriptions had left them confused and not knowing who to trust or what to believe.
I won’t forget April’s dad’s shout of recognition. By giving them the same diagnosis as AI, I had instantly cemented their confidence in this being the right explanation. AI is not going to replace the need for human clinicians, but it clearly has the potential to be a useful ally in the quest to understand and interpret when something goes wrong with our health.
[Further reading: A forgotten theory that explains Trump’s psychosis]
Content from our partners
Related
This article appears in the 22 Apr 2026 issue of the New Statesman, All alone
Join the debate
Subscribe here to comment