The rapid evolution of the coronavirus into an alphabet soup of subvariants presents a vexing challenge to health officials: They must make far-reaching policy decisions based on little biological certainty of which viral variants will be dominant this fall or winter.

The Food and Drug Administration said at the end of June that it would update coronavirus vaccines for a booster campaign in the fall targeting highly contagious Omicron subvariants. But the ground is shifting beneath its feet.

In just eight weeks, the BA.5 subvariant has gone from a blip in U.S. case counts to the dominant subvariant in the country, now making up more than three-quarters of new cases. Perhaps the most transmissible subvariant yet, it is pushing up positive tests, hospitalizations and intensive care admissions across the country,

There is no evidence that BA.5 causes more severe disease, but the latest metrics certainly bust the myth that the virus will become milder as it evolves.

“None of us has a crystal ball, and we are trying to use every last ounce of what we can from predictive modeling and from the data that we have to try to get ahead of a virus that has been very crafty,” said Dr. Peter Marks, a top vaccine regulator at the F.D.A., after an advisory committee recommended that the agency prioritize vaccines specific to the Omicron subvariants. “For something that’s only nanometers in size, it’s pretty darn crafty. We’re trying to make our best judgment here.”

Omicron and its offshoots have dominated for about six months now, and whatever incarnation of the virus comes next is more likely to be tied to the Omicron family than to earlier versions, said Jerry Weir, a senior F.D.A. regulator.

That assumption is the best calculation that can be made at this time, according to outside experts not on the F.D.A. expert panel.

“Viruses like SARS-CoV-2 are always evolving, and it’s a near certainty that new mutants will emerge in any given six-month time frame,” said Jesse Bloom of the Fred Hutchinson Cancer Center in Seattle. “But as long as these mutants are descendants or close relatives of BA.2 or BA.4/BA.5, then a vaccine booster based on BA.4/BA.5, as the F.D.A. has recommended, should be a much better match to them than the current vaccine, even if it’s not a perfect match.”

The Centers for Disease Control and Prevention on Tuesday cleared a fourth vaccine, from Novavax, for use in the United States, but its trials were conducted before the emergence of Omicron and its effectiveness against the variant could be limited.

The Omicron family tree has been growing faster since Omicron BA.1 was first detected in late November 2021. New federal estimates on Tuesday showed that BA.5 made up about 78 percent of new cases in the United States, as of last week.

Emma Hodcroft, a molecular epidemiologist and researcher at the University of Bern in Switzerland, said that Omicron’s evolution pattern has diverged from earlier variants. “The children of Delta were not dominant, but the children of Omicron are pushing out their siblings, if you will,” she said. “That’s hinting that Omicron is at the peak and that there will be smaller changes.”

Though more Omicron children could be on the way, she and other scientists emphasized that would not preclude another variant appearing.

“Too many times we have made predictions on how we think SARS-CoV-2 will evolve and then been emphatically wrong,” said Nathan Grubaugh, an epidemiologist at the Yale School of Public Health. Last fall, he correctly predicted that an immune-evasive subvariant would emerge, but his expectation that it would come from the Delta variant was wrong.

“Obviously, we are seeing new variants emerge from within Omicron right now — BA.2, BA.4, and BA.5 — and that may continue to happen,” he said. “But we shouldn’t become unimaginative and think that will continue.”

Last year, ​​Sarah Cobey, an evolutionary biologist at the University of Chicago, was almost certain that the next variant would descend from Delta. “I nonetheless think it’s extremely likely the next variant will descend from Omicron,” she said last week, adding that it could have a wider degree of immune escape or increased transmissibility. “It’s likely the next variant has emerged already but will evade surveillance for some time,” she said.

An ominous Omicron cousin, BA.2.75, is already being seen in parts of the world.

That makes tweaking the formula for booster shots more critical, according to Kristian Andersen, a virologist at the Scripps Research Institute in La Jolla, Calif. The benefits of such a move have much more to do with broadening immunity than with increasing protection against a very specific variant, he said. Early variants did not evolve from previous variants but rather from earlier lineages, he said, unlike Omicron, making predictions harder.

Forecasting, though, is every expert’s stock in trade.

“We’ve been trying to better guess the next flu variant for decades,” Dr. Hodcroft said. “And it turns out that it’s very complicated.”

The many variables mean that she and other experts cannot make a declaration with full confidence. She said: “It’s very hard to put these all into a machine and crunch it.”