Key Takeaways from IDWeek 2025: AI, AMR Strategies, Next-Generation Anti-Viral Development Converging

IDWeek 2025 marked a pivotal moment in the evolution of infectious diseases (ID) research and pharmaceutical innovation. Beyond its traditional focus on clinical management and epidemiology, this year’s conference spotlighted the accelerating convergence of artificial intelligence (AI), antimicrobial resistance (AMR) strategy, and next-generation anti-viral development. From AI-driven drug discovery platforms to real-world evidence (RWE) reshaping regulatory and market access pathways, the tone was clear:  ID therapeutics are no longer a reactive field—they are becoming a proactive, technology-enabled ecosystem central to global health preparedness and sustainable drug development.

Below, Medpace ID&V experts highlight the top takeaways from IDWeek 2025. Read on to explore key themes and advancements.

Key Conference Takeaways

1. AMR drug development problem is still an R&D opportunity, not just a stewardship problem.  

IDWeek 2025 leaned into AMR drug development but with a noticeable focus on “pipeline and commercialization”. Multiple sessions focused on using RWE to justify earlier use of newer Gram-negative agents and targeting multidrug-resistant organisms in high-risk, complex patients (transplant, ICU, ventilated pneumonia, etc.)—and not just for salvage use.

Shionogi presented new surveillance and real-world outcomes data on cefiderocol, a siderophore cephalosporin for highly resistant Gram-negative infections. The take home message was that if you move cefiderocol earlier in the treatment sequence instead of holding it in reserve, you get better outcomes in difficult-to-treat infections. This is important for drug development for several reasons:

• Payors and hospital P&T committees are still the main bottleneck for novel anti-infectives. High quality RWE is being positioned as an alternative to massive, traditional Phase 3 trials for market uptake in severe, resistant infections.
• Regulators (largely missing from this year’s meeting due to the government shutdown) have already signaled openness to pathogen-focused and site-agnostic trial designs in AMR trials. The field is now adding pragmatic post-market data to argue for broader label use and reimbursement sooner.

The key takeaway here is that AMR companies showed up at IDWeek 2025 to not just say “we kill XDR Acinetobacter,” but to say, “here is compelling RWE to support guidelines and earlier empiric use of novel antibacterials.” That’s a commercial strategy shift, and it affects how AMR assets are being developed and de-risked—hopefully a message that will encourage stakeholders in AMR R&D to keep the fire lit.

2. Broad-spectrum direct-acting antivirals are trying to be “the next Tamiflu—but for everything!”

One of the widely discussed preclinical stories was MDL-001, described as an orally available, direct-acting, broad-spectrum antiviral. The drug was developed from an array of AI models within Model Medicines’ proprietary platform, targets a conserved “Thumb-1” domain in viral polymerases, and was presented as a novel, cross-family antiviral mechanism validated across respiratory and hepatic viruses. The team positioning it (academic and biotech collaborations including UCSD, Scripps, and Mount Sinai) framed MDL-001 as potentially the first in-class small molecule that could be deployed broadly across multiple, unrelated viral infections. Why this matters:

• The COVID era proved that governments will stockpile pan-respiratory countermeasures. If you can credibly pitch one oral agent as deployable for multiple viral threats (flu, RSV-like viruses, coronaviruses, and hepatic viruses), you become a biodefense-pandemic-preparedness asset, not just a single-indication antiviral.
• Mechanistically, going after a deeply conserved viral polymerase motif is a key development strategy that goes beyond the classic, single-virus, single-protein model of antiviral drug development.
• These broad-spectrum programs are increasingly using AI-driven comparative virology and protein structure modeling to identify conserved pockets that are both druggable and mutation-resistant. That is the same logic that was driving AI-designed pan-variant SARS-CoV-2 inhibitors over the last 2-3 years, now applied to multi-family antivirals.
• This has pipeline strategy implications. We’re witnessing the emergence of a new approach to antiviral drug development, and IDWeek 2025 gave this approach the spotlight rather than leaving it confined to translational journals.

3. AI has moved from hype to actual workflows in infectious diseases R&D.

AI wasn’t just a keynote talking point at IDWeek 2025—it had dedicated sessions for clinicians, epidemiologists, and translational researchers.

How AI is enabling translational research and drug development:

• Target discovery and hit finding. Researchers highlighted AI multi-agent systems that can:
  • Mine pathogen genomes and resistance plasmids for novel essential targets
  • Generate first-pass inhibitor scaffolds
  • Evaluate pharmacologic liabilities (PK, tox, CMC) in silico before any wet-lab spend
  • Drug hunters and drug discoverers are using these AI tools to prioritize anti-infectives that hit resistance-proof bacterial pathways and antivirals that bind pockets that are slow to mutate. This may mean that there are “fewer shots on goal,” but each shot is smarter and cheaper, which is crucial in ID where traditional ROI is weak.
• Trial design and site selection. Speakers described using AI and large language models (LLMs) on top of hospital EHR data to:
  • Predict which hospitals will actually enroll the right resistant isolates (Where will 4+ antibacterial class drug-resistant Pseudomonas be seen in the first half of next year?)
  • Simulate inclusion/exclusion criteria against historical patient populations to avoid overly restrictive trials
  • Streamline workflows and support expedited document generation
  • Generate adaptive trial schemes tuned to the dynamics of outbreaks rather than static timelines developed months before a study actually starts
  • This can shorten the path from preclinical “cool mechanism” to first-in-human and then to pivotal data, as you are less likely to design an underpowered, un-enrollable ID trial.
• Real-time resistance surveillance as a development input. IDWeek 2025 programming around hospital epidemiology and stewardship emphasized AI-driven resistance dashboards: models that continuously learn from antibiograms, culture data, and prescribing patterns to flag emerging resistance clusters inside and across institutions. This is not just for infection control. It’s critical knowledge for pharma.
  • You can watch resistance drift in near-real-time and aim your next gen β-lactamase inhibitor or siderophore cephalosporin exactly where the cracks are forming.
  • This dynamic “heat map of unmet need” is being fed back upstream into medicinal chemistry prioritization. In other words, hospital surveillance data is now a discovery tool, not just a stewardship tool.

4. Vaccines and preventives: durability, access, and combination strategies.

Another repeated theme at IDWeek 2025 was the durability and positioning of adult respiratory vaccines (RSV, flu, COVID) and shingles vaccination in older adults. Presenters previewed updated effectiveness data and adherence data in real-world settings. Why this matters for developers:

• Respiratory vaccines are being repositioned from “seasonal public health tools” to “chronic risk modifiers in older patients with comorbidities,” which supports combination products, long-acting monoclonals, and year-round prophylaxis strategies.
• This kind of framing (lifetime risk management instead of outbreak response) changes how companies think about trial endpoints and commercial lifetime value.

Again, AI shows up here in immunogen design and epitope mapping. Structural vaccinology has already leaned on protein-folding AI, but what’s new is the LLM-style models trained on immunology literature that are now being used to propose which epitopes are most likely to drive durable neutralization without antibody-dependent enhancement (ADE) surprises. This was referenced in discussion of optimizing next-gen RSV and pan-flu immunogens.

5. Public health stressors are directly reshaping ID clinical trials.

A big, uncomfortable subtext in Atlanta: federal participation was partially disrupted because of the US government shutdown, which limited CDC scientists’ ability to attend and present even when they were on the printed program.

Why that’s not just politics:

• The CDC normally supplies critical epidemiology, genomic surveillance, and outbreak analytics that feed biotech pipeline decisions (Which clone of carbapenem-resistant Klebsiella is spreading? Which H5N1 clade looks like a spillover risk?)
• With limited to no CDC voices in the room, you could sense industry and academic groups stepping in to their own surveillance, modelling, and preparedness narratives.

That shift matters for drug development:

• When biopharma and academics start doing the epidemiology that traditionally lived inside federal agencies, they unofficially also start setting the “priority threat list.” The assets they’re already building naturally bubble to the top of that list.
• You can think of this as privatized prioritization. That accelerates certain antiviral/antibacterial programs and may starve others. The odds are better for the best-funded and loudest voices.

What’s Next for ID Drug Development After IDWeek 2025

Following IDWeek 2025, the trajectory of ID&V drug development appears to be shifting. AI-native discovery and triage are no longer theoretical—multi-agent LLM systems are being openly taught to clinicians, which design, score, and iteratively improve leads for resistant organisms and emerging threats before a single mouse study is run. Additionally, pan-pathogen/pan-family antivirals are real programs with named molecules that will move forward.

Meanwhile, AMR economics are being attacked with RWE, not just policy lobbying. Companies are generating outcomes data that argue for earlier-line use of advanced agents, reframing them from niche “salvage drugs” to “standard tools” to prevent mortality in high-risk patients. That’s how you justify premium pricing and formulary inclusion, which in turn justifies keeping AMR discovery alive.

Clinical trial ops are getting “precision-ID”. Generative AI is being used to pre-map enrollment sites, predict resistant isolate flow, and auto-generate adaptive trail designs. This could cut months (and millions of dollars) between preclinical signal and registrational data in ID—historically a slow, expensive area to prove benefit because events are sporadic and resistant patterns move. Finally, preparedness/biodefense framing is now backed into mainstream ID pipelines. Between pan-respiratory antivirals, adult respiratory vaccine durability data, and climate-driven vector-borne threat sessions, the pitch is: “These are not niche outbreak tools. These are infrastructure.”

Highlights for Drug Developers and Investors

IDWeek 2025 in Atlanta was focused less on this year’s bug of concern and more on rebuilding the ID pipeline with AI as the engine, RWE as the validator, and pandemic preparedness as the business model. The throughline is that ID therapeutics—long considered underfunded and low-margin—are being reframed as critical infrastructure with faster, cheaper discovery cycles and broader commercial justification.

Unable to connect with us during IDWeek? We welcome the opportunity to discuss your upcoming clinical development plans. Contact Medpace ID&V experts today.

An Industry Leader in ID&V Clinical Development

Medpace’s specialized cross-functional team of experts have extensive experience executing global trials in antivirals, antibacterials, antifungals, vaccines, and immunotherapies spanning adult and pediatric patients, as well as special at-risk populations. Our experienced medical leadership team, integrated laboratories and clinical trial management system, and extensive site and investigator relationships produce a truly seamless drug development process