MIT team uses an AI algorithm to screen excipient ratios, yielding RNA vaccines that stay stable for a year at room temperature or two months at 37 C and still match a Moderna-like vaccine's immune response in mice

The team used an AI algorithm to predict excipient ratios, compressing formulation screening that would otherwise require extensive experimentation into a few weeks and arriving at an LNP formulation with markedly improved heat stability. The same group had previously developed polymer-stabilized LNPs that withstand higher temperatures, but those LNPs differed somewhat from the FDA-approved formulations used in the Moderna and Pfizer Covid-19 vaccines; this work targets those approved formulations directly. The text details the workflow: nearly 50 FDA-approved excipients were each measured for how well they stabilized RNA, five of the most promising were selected, the algorithm predicted ratios, two formulations at a time were tested in cells with results fed back into the algorithm, and after several rounds one formulation advanced to animal studies; the authors report this took only a few weeks, whereas months of manual screening previously yielded that 'nothing would get us to 100 percent stability.'

The reformulated LNPs still induce immune responses comparable to the comparator vaccine after prolonged storage at high temperature and at room temperature. Conventional RNA-LNP vaccines require storage at -20 to -80 C; this work relaxes storage to two months at 37 C or one year at room temperature without losing immune performance. Covid-19 mRNA antigens were packaged in the particles, dehydrated by vacuum drying, stored under those conditions, and vaccinated mice showed immune responses equivalent to mice receiving vaccines carried by LNPs similar to the original Moderna formulation; this is a mouse-level equivalence result.

The same heat-resistant formulation can be made into solid microneedle patches delivering a SARS-CoV-2 antigen and produces immune responses similar to injectable RNA vaccines. It pairs the heat-stable formulation with microneedle patch delivery, pointing toward vaccination without cold chain or injection. The text reports that the patches generated an immune response similar to that produced by the injectable RNA vaccines; this is an animal-level immunogenicity comparison.

The algorithm transfers to other LNP formulations, including one similar to the Pfizer Covid-19 vaccine's, and a given heat-resistant formulation could in principle carry any mRNA payload. This indicates the approach is a reusable formulation-stabilization workflow rather than one-off tuning for a single vaccine. The authors demonstrate stabilization of a Pfizer-like LNP that uses the same excipients as the Moderna-oriented formulation but in a different ratio; the claim about adapting to any mRNA payload is a forward-looking statement by the researchers rather than a result measured in this study.

This work addresses RNA vaccines and delivery platforms that need to move beyond ultracold storage: it fits formulation development that wants to keep the approved Moderna- and Pfizer-style LNP systems while gaining higher heat stability, and it fits delivery formats such as microneedle patches and controlled-release particles that require a solid state or tolerance of higher temperatures. Beneficiaries include vaccine developers, delivery-system researchers, and immunization programs that depend on distribution in cold-chain-limited settings. Methodologically, it suits formulation optimization where only small datasets and limited experiment counts are available.

Heat resistance and immune-response equivalence currently come from cell and mouse experiments, and performance in humans still needs follow-up study. The one-year room-temperature and two-month 37 C stability were measured under vacuum-dried (solid) conditions, and how this maps to liquid formulations is not clear. The algorithm operated within a set of nearly 50 FDA-approved excipients, and behavior beyond that excipient set remains to be seen. The claim that a formulation could carry any mRNA payload is a researcher outlook that this paper does not verify one by one. In addition, this is a public-facing news account that does not give the specific formulation composition, statistical details, or figure data, so the precise range of effect sizes cannot be judged from the text.