Title

Presentation

Spectral AI, Inc. beats earnings expectations. Reported EPS is $-0.06, expectations were $-0.13.

Operator: Good afternoon, and welcome to the Spectral AI Fourth Quarter and Full Year 2024 Financial Results Conference Call. All participants will be in listen-only mode. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Sara Prendergast, Assistant General Counsel. Please go ahead.

Sara Prendergast: Thank you, Gary. Good afternoon, everyone, and thank you for joining us for Spectral AI’s 2024 fourth quarter and full year financial results conference call. Our speakers for today will be Dr. Michael DiMaio, MD, Chairman of the Board of Spectral AI; and Vince Capone, the company’s Chief Financial Officer. We also have Dr. Jeffrey Carter, MD, who is the principal investigator in our Burn validation study and Director of Trauma and Burn at the University Medical Center, New Orleans. And Dr. Jihang Wang, the company’s Director of Data Science with us today. Before we begin, I’d like to remind everyone that during this call, certain statements made in this release are forward-looking statements within the meaning of the Safe Harbor provisions of the United States Private Securities Litigation Reform Act of 1995, including statements regarding the company’s strategy, plans, objectives, initiatives and financial outlook.

When used in this press release, the words estimates, projected, expects, anticipates, forecasts, plans, intends, believes, seeks, may, will, should, future, propose, and variations of these words or similar expressions, or the negative versions of such words or expressions, are intended to identify forward-looking statements. These forward-looking statements are not guarantees of future performance, conditions, or results, and involve a number of known and unknown risks, uncertainties, assumptions and other important factors, many of which are outside the company’s control that could cause actual results or outcomes to differ materially from those discussed in the forward-looking statements. As such, listeners are cautioned not to place undue reliance on any forward-looking statements.

Investors should carefully consider the foregoing factors and the other risks and uncertainties described in the Risk Factors sections of the company’s filings with the SEC, including the registration statement and the other documents filed by the company. These filings identify and address other important risks and uncertainties that could cause actual events and results to differ materially from those contained in the forward-looking statements. Management will also be discussing non-GAAP financial metrics. A reconciliation of these non-GAAP financial measures to the most comparable GAAP measures can be found in the company’s press release. With that said, I’d now like to turn the call over to Dr. DiMaio. Dr. DiMaio, please go ahead.

Michael DiMaio: Thank you very much, Sara, and good afternoon, everyone. We very much appreciate you joining today’s call to discuss our fourth quarter and full year 2024 results and provide you an update on our progress. 2024 was a pivotal year for Spectral AI as we continued advancing our DeepView technology to transform the assessment and potential treatment of wounds and burns. Our efforts over the last year have positioned us for long-term success, marked by key milestones in our regulatory pathway, clinical validation and strategic partnerships. For those of you who don’t know, I am the founder of Spectral AI. We founded the company as SpectralMD in 2009, and I was at Spectral AI’s — I Am, the second largest shareholder.

I once served as CEO, getting the BARDA contract and other matters, and returned to the role of Chairman of the Board last October. In our third quarter earnings call, I said we would get back to basics, and I said we would focus on the three F’s, finance, focus, and finish. Our efforts are indeed paying off. And earlier this month, we announced positive top line results of our Burn validation study, which was led by Dr. Jeff Carter from New Orleans. He demonstrate the exceptional performance of our DeepView system in predicting burn healing potential. This study, one of the largest burn trials ever conducted in the U.S., enrolled 164 patients for a potential de novo application across both adult and pediatric populations. The findings confirm that DeepView significantly outperforms the clinical judgment of burn physicians in three key diagnostic metrics, sensitivity, diet score, and specificity, which were all reviewed with the FDA.

For sensitivity, DeepView identified non-healing tissue with over twice the accuracy of burn physicians, achieving 86.6% at the image level compared to 40.8% from clinicians’ judgment. The second measurement, Dice Score, DeepView system segmentation accuracy in AI algorithm reached 68.5% compared to 39.2% for the burn physicians. Finally, specificity, the DeepView system surpassed anticipated results in identifying non-healing wound areas, achieving 61.2% versus the expected 36%. These results reinforce DeepView’s potential as a transformative tool in burn care. Our goals are to provide clinicians with an immediate, objective assessment on day one of an injury, improving patient outcomes, and potentially reducing health care costs. We plan to submit these results to the FDA by the end of Q2 2025, seeking De Novo classification in 2026.

We are very proud of this milestone due to the focus that we’ve achieved on delivering the items that we promised to BARDA, the U.S Military, and the population at large. We are now actively gathering real-world clinical feedback across two continents, further validating the system’s impact. These global evaluations will provide critical insights as we prepare for our upcoming FDA submission and eventual commercialization in the U.S. and beyond. In the U.K. and Australia, the response to our system has been exceptional. Burn [ph] specialists there have expressed strong enthusiasm for the technology, recognizing its ability to improve patient care. We are committed to expanding our presence in the U.K and further integrating DeepView into standard burn care protocols.

At the end of our third quarter 2024 conference call, I was asked if I could provide a timeline for bringing DeepView to Australia. At the time, I stated that we expected to have devices prepared and rolled out within the next few months with more details to share on this call. I am happy to report that we have delivered on that commitment. The momentum surrounding our technology has extended beyond the U.K with partners in Australia actively integrating DeepView into their hospitals. We successfully deployed three DeepView systems in Australia at leading hospitals, including one, Concord Repatriation General Hospital in Sydney; two, the Royal Adelaide Hospital in Adelaide; and third, Fiona Stanley Hospital in Murdoch, which is a suburb of Perth.

These deployments were made possible through Australia’s special access scheme in collaboration with PolyNovo Limited, a global leader in burn wound therapy. By expanding into the U.K and Australia, we are gathering valuable real-world clinical data that will strengthen our case for our FDA submission and support global commercialization strategies across the U.S. and beyond. In terms of DeepView SnapShot M, this is a fully handheld, portable and wireless diagnostic tool built on the same AI-powered platform as a regular DeepView system. It is designed to provide rapid, real-time wound assessment in military and emergency care settings, including battlefield triage, forward-deployed medical units, ambulances and home health care environments.

The development of SnapShot M has been heavily supported by government funding, with total awards now exceeding $7 million. These awards include multiple grants from the Department of Defense, the Defense Health Agency, and MTEC, the Medical Technology Enterprise Consortium, which most recently awarded an additional $850,000 to further its development. We anticipate SnapShot M to be ready for military deployment in 2027 with potential applications beyond combat medicine, including wound size measurement and expanded emergency and home health care use. These investments reinforce a broad potential of DeepView technology, allowing us to address critical needs across multiple sectors. Discussing Sim IP. In March 2024, we announced the formation of Spectral IP, a dedicated intellectual property subsidiary focused on monetizing IP assets within the health care and AI-driven technology sectors.

Shortly after its formation, Spectral IP secured a $1 million investment with Erich Spangenberg, who is a leading figure in global IP investment and Spectral AI’s largest shareholder, appointed as its Chief Executive Officer. In November 2024, we took the next step by announcing our intention to spin off Spectral IP into an independent publicly traded company through a transaction with Sauvegarder Investment Management. The following, the spin off, Sim IP Incorporated, will focus on IP-based financing and monetization strategies including licensing, litigation investment, structured financing, royalty acquisitions, and opportunistic transactions. While industry agnostic, SIM IP will prioritize health care, semiconductors, AI and IoT sectors where intellectual property remains and underutilized via valuable asset class.

Spectral AI’s 30 patents and 35 pending patent applications will remain with Spectral AI and not Sim IP. The spin-off is expected to be executed through a distribution of Spectral IP stock to Spectral AI shareholders subject to final Board and repertory approvals and other closing conditions. While we anticipate completing this transaction within the next 90 days, there can be no assurance regarding the ultimate timing or completion of this transaction. With that, I will turn it over to our Chief Financial Officer, Vince Capone. Vince?

Vincent Capone: Thanks, Dr. DiMaio, and thank you all for joining us today. I would like to remind everyone that our press release issued this afternoon contains additional details on our operating results. We expect to file our Form 10-K with the SEC later this week. With that in mind, I will focus my remarks on select financial highlights and key items. Starting with our fourth quarter results, research and development revenue was $7.6 million as compared to $5.3 million in the prior year period. This increase reflects the additional work we’ve done in the fourth quarter under the BARDA PBS contract. As a reminder, especially for those of you who may be new to Spectral, R&D revenue is driven by our research and development activities.

We invoice the U.S. government monthly to reflect our R&D expenditures with an added margin to cover expenses such as labor and third-party contractors and consultants. We generally receive payment against these invoices from the U.S. government shortly after our invoice submission, which produces a positive and predictable effect on our P&L and our cash flow. Gross margin in the fourth quarter declined to 44% as compared to 46.1% in the prior year period, due primarily to a smaller percentage of direct labor as a proportion of the total work on the BARDA PBS contract for the quarter. General and administrative expenses in the fourth quarter decreased to $4.6 million as compared to $5.4 million in the prior year period, driven by a decrease in our non-BARDA related development activities from 2023.

Moving on to our full year results, I am proud to announce that our research and development revenue in 2024 was $29.6 million, a 63.5% increase from the $18.1 million we recorded in 2023, representing the highest revenue reported in the company’s history. The year-over-year increase reflects the expanded work on the BARDA PBS contract for the full year 2024. Gross margin rose to 44.9% from 43.6% in the prior year period, driven by the higher reimbursement rate under the current BARDA PBS contract, as compared to the BARDA Burn II contract, which was in place for most of 2023, therefore creating a positive impact on the company’s gross margin. General and administrative expenses in 2024 was $19.9 million, as compared to $20.9 million in 2023, primarily driven by a decrease in the non-reimbursed work and the higher legal and accounting spend that we incurred in 2023 relating to the company’s financial transactions.

As a component of our general and administrative expenses, R&D expenses in 2024 increased to $19.3 million from $15.1 million in 2023. It’s important to note that total R&D activity is reflected both in our general and administrative expense and in our cost of revenue. As a result, the increase in 2024 is primarily driven again by the increased work on our BARDA PBS contract. As of December 31, 2024, cash and cash equivalents totaled $5.2 million. Following the completion of the long-term debt financing with Avenue Capital, which we announced earlier this week, I’m pleased to announce that our cash position currently exceeds over $14 million. Regarding our liquidity, I want to highlight the additional following items. Total monthly operating losses were reduced from $4.6 million for the first half of 2024 to approximately $2 million in the second half of 2024 as we focused on improving our operating efficiencies and the concentration on the DeepView system as part of the BARDA PBS contract.

In 2024, we were also able to satisfy approximately $10.1 million of our short-term debt with an institutional lender, which was fully satisfied in February 2025. We believe that our recent long-term debt financing agreement and the smaller equity raise completed in connection therewith, in combination with the multi-year non-dilutive funding provided by our U.S government contracts with approximately $113 million left to utilize, provides us with a sound financial foundation to pursue our strategic objectives and bring our DeepView system to commercialization. The company is forecasting revenue of approximately $21.5 million for fiscal year 2025, and the financial guidance for fiscal year 2025 does not reflect any contributions from the sale of DeepView system for burn in the U.K or in Australia or any additional material financial contributions that may result from the commercialization of our DeepView system.

With that, I will turn it back over to Dr. DiMaio.

Michael DiMaio: Thank you, Vince. Before turning to your questions, I want to thank our team for their focus and dedication and commitment to our promise to develop and commercialize our DeepView system. As I said, when I took over as Chairman, I wanted to focus on three F’s I mentioned earlier, and the team has responded very well to that motivation. Their achievements to date and those on the horizon drive our success. The management team has coalesced around these goals and are excited to leverage the sound financial foundation that Vince has described in supporting the team’s efforts. Operator, let’s open the call for questions from the analyst, please.

Operator: [Operator Instructions] Our first question today is from Carl Byrnes with Northland Capital Markets. Please go ahead.

Carl Byrnes: Thanks. Congratulations on your progress. I’m wondering what your thoughts are, or your plans with respect to commercial readiness and the launch, obviously, around the corner with respect to staffing, regions, if you’re contemplating any co-promotion partners, and the like. Thanks.

Michael DiMaio: Thank you. Great question, Gary. So, if I heard you say that what are our plans for commercialization, so let me go a half step back, of course. We’re going to anticipate the FDA clearance with the Dendrodome submission. That should yield a commercially viable product that we will begin to commercialize. We’re going to precondition the market with a number of ways. There are a number of studies that we’re going to be utilizing in that time period between the submission and the clearance, the De Novo submission. We have presentations, papers, publications, et cetera, in addition to mobilizing a sales force that will begin to accumulate as well. But as I mentioned in the discussion, with the devices in the U.K and Australia, we’re getting RWE, or real-world evidence, to support the devices in those countries can support and help the care of the patients in those countries.

Therefore, we plan on abstracts and writing and other matters to be able to discuss that to be able to have the excitement around commercializing it more broadly. As you asked as well about other matters and co-products and all that, that’s currently being discussed, but I can’t report anything at this time that would be material to disclose.

Carl Byrnes: Got it. Thanks. That’s helpful. I’m wondering, though, kind of going back, if you were to, on a complete go alone strategy, what would you envision in terms of how many sales reps would you need and how many medical specialists would you need or liaisons in regions? Thanks.

Michael DiMaio: And again, another good question. We already have a number of service people out there taking care of the devices that are out there for the research purposes of the validation study, I would be — I don’t want to stretch out and say exactly how many I would have at this time. That would be premature, but we’re looking at approximately 170 more units being rolled out by the federal government contract in the PBS contract. So, we would need the appropriate number at this time, which I wouldn’t be free to give at this time. There’s no sales force necessarily, because the BARDA — the devices that BARDA plans to roll out will not need a sales force to roll those out. So it would be beginning small sales force necessary to get those initial 170 units to the U.S.

Carl Byrnes: Got it. Thanks.

Michael DiMaio: Thank you.

Vincent Capone: Thanks, Carl.

Operator: The next question is from Ryan Zimmerman with BTIG. Please go ahead.

Ryan Zimmerman: Hey, guys, can you hear me okay?

Michael DiMaio: We can hear you great.

Ryan Zimmerman: Okay, good. Sorry for the background noise. Santa Airport. Maybe on the $21 million in BARDA R&D revenue this year for 2025, I think before we thought about it kind of similar to 2024, Vince, just kind of bridge us from what was the $28 million guidance in ’24 to the ’21 in 2025. And then, with that and not including any commercial revenues, but then the comment that you made that you do expect commercial revenue potentially this year, just want to understand kind of how you’re thinking about your commercial launch in terms of contributions.

Michael DiMaio: Yes, Ryan, that’s — I mean it’s a good question here. I mean, we have always seen 2025 as a little step back in revenue as we position for the FDA submission. And, obviously, we can’t roll out in the U.S. until we get the de novo clearance. So 2025 will be a step back in revenue, but we’re really going to be in the process of providing additional developments to the device so that when we get clearance from the FDA, we’ll be in a position to really reach commercialization very quickly for the BARDA contract first and then for other indications throughout. Again, and we talked about this in earlier calls, revenue generating from the U.K and or Australia, while it would be nice to see, it’s not going to be dispositive for us as far as a significant portion of overall revenue, but we’ve only seen 2025 as kind of a little retreat from 2024 as we pause to get commercialization and FDA clearance and we look to see that ramp up in 2026.

Ryan Zimmerman: Okay. And then Dr. DiMaio, what’s left to submit the FDA [ph], the results were really good. They were really encouraging a few weeks ago. What do you got to do? What other boxes do we have to check just to get that final submission in?

Michael DiMaio: It’s all the things you might expect. It’s the SAP, the statistical plan, the — all the hardware and software requirements, the validation and V&V that goes along with all that, the clear results of the clinical trial that I described the results to, it’s an entire package of how many sections would that be? 14 sections for the FDA application. I might also add to that, because of our BARDA relationship and other U.S. government agency, all of the sections will be previewed and reviewed by BARDA first, and then go to the FDA. So we’ll have another set of eyes, if you will, and brains reviewing everything before it goes into the FDA.

Ryan Zimmerman: Thank you.

Operator: The next question is from John Vandermosten with Zacks. Please go ahead.

John Vandermosten: Great. Thank you and hello, Dr. DiMaio and Vince. You released the data a short time ago, and there was the Dice Score mentioned. And I wanted to see if you could give us some additional context in terms of what that means and how the FDA looks at it. And then also the importance in measuring at the image level versus the pixel level.

Michael DiMaio: John, that is a fantastic question and anticipating that coming. I have two of our experts here, Dr. Wong — Jihang Wong, who’s our Data Science Chief, and Dr. Jeffrey Carter, who’s run the PI trial. So I’m going to turn it over to Dr. Wang to give us a couple of answers on your questions, which are excellent.

Jihang Wang: Thanks, Dr. DiMaio. So, John, this is a good question. So for Dice, basically, it’s a widely used metric in data science field for evaluating the segmentation path, including non-helium segmentation in our case. Basically, it measures the overlap between the DEI model segmentation and versus the ground truth of the non-healing region. The benefits I feel that FDA requires Dice here is because FDA not only wants to understand our algorithm’s sensitivity, but it also wants to understand the Dice Score, which is kind of like a balance, which will provide some balance results for both the precision and recall. Because if you know sensitivity, basically, it will provide you how good you actually segmenting a target, but it won’t penalize if you the algorithm makes some mistakes or have some false positive.

Well, Dice actually tried to balancing both the precision and recall, which actually gives a more comprehensive view of the models performance, especially in the cases of the imbalanced data sets. And regarding your next question here, during the discussion with the FDA, I think when we actually report our segmentation performance, it’s also a natural reporting about its detection performance because if you think about segmentation, it’s a more advanced way to report the results, right? You get a contour of the burn. But I think the first step FDA wants to understand is whether your device is able to actually identify the non-healing area in the image. So that’s why, for FDA’s suggestion, we have the two paths of the validation process, one focused on the detection performance and another one focused on the segmentation performance.

John Vandermosten: Okay, great. Another question is on additional deployments in Australia and the U.K. You mentioned there are three in Australia. What is on tap for 2025 for the rest of the year there? Do we think we are going to keep up that pace, or — yes just wondering what the pace might be for the rest of the year.

Michael DiMaio: We don’t plan on putting in any more units, if that is your question, in terms of pace. At those three centers, the pace would be that they have been trained to begin to utilize it, just like the U.K physicians begin to understand how to utilize it, when to utilize it, and give us the feedback that we look forward to with a real-world application. So we don’t expect to expand the number of sites. We just expect to do a deeper dive in each of them to understand better how it’s used. And as we make serial improvements, which we expect to make with the feedback, how we loop that back in.

John Vandermosten: Okay. So we have the expected number of units both in the U.K and Australia right now?

Michael DiMaio: Yes, sir. That is correct.

John Vandermosten: That is the foreseeable future. Okay. Thank you.

Michael DiMaio: Thank you.

Vincent Capone: Thanks, John.

Operator: The next question is from Ramakanth Swayampakula with H.C. Wainwright. Please go ahead.

Ramakanth Swayampakula: Thank you. Good afternoon, Michael and Vince. Couple of quick questions. So in terms of the pediatric study that got initiated recently, what’s the timeline for data from that study? And how would you use that data to expand your label?

Michael DiMaio: Hey, R.K., thank you very much. The data on the study has been — I discussed it in this call, and we also had an issue of a press release a week or so ago, two weeks ago, with the top line numbers. We hit the performance metrics that were asked and cleared by the FDA. So the answer to your question is the data has been preliminarily processed and true, and that’s what we have planned to release. That’s locked down. I can repeat the numbers if you want.

Ramakanth Swayampakula: No, I’m talking about the pediatric study. It’s all, our case, it’s all merged into one. There’s no separate studies. Both the burn centers and emergency department centers merge, as well as the pediatric and adult populations. It’s all merged into one big data set.

Michael DiMaio: Does that answer R.K.? I mean, we may do some analysis. It’s not the requirement of the FDA.

Ramakanth Swayampakula: I didn’t see that press release. Okay.

Michael DiMaio: Yes, I’m sorry. Yes. That’s a good question, though, because that we did, I’m sorry to interrupt you, we did enroll 115 adults and 49 pediatric patients for a total of 164. So we feel we have a representative sample to, again, merge all this data. And to repeat what I said, we’ll be doing some analyses, but that is not the requirement that we anticipate for the FDA submission that we will submit.

Vincent Capone: Okay. And then in terms of the DPU Snap Shot M, is there any milestones that we can expect in terms of how the development is going during ’25? So there are some soft milestones. Actually, the team went down to the San Antonio Burn Center, the military, got some preliminary feedback from the military down there. Those are being incorporated right now. We’ll be submitting some of those plans to the remainder of 2025. So this is kind of a completion, I would say, in 2025 to get that to the point of a more formal, I want to call it clearance, but we’re looking at ruggedness, testing, what other changes we need to make based upon the military using it in the preliminary assessment. In 2026, I believe it is, I can’t recall the date right now, we’ll be sending in another application I think it’s in March of 2026 through September of next year, getting the next tranche of money from the federal government or the military to be able to complete the processing.

Of course, what the military wants is for us to be able to design and develop it to be able to use in more austere settings. Again, as we know, the world is a very difficult place, and so higher altitudes and temperature extremes, all of that will be further developed with the device.

Ramakanth Swayampakula: Okay. And then in terms of the product that has been placed in U.K., it’s been, I think it’s been a little over a year now. So what are the learnings from that? And from what you’re learning from there, do you think there has to be any protocol changes or any development on the software or hardware side?

Michael DiMaio: R.K., let me turn over to Dr. Carter who is our burn expert and has been watching all that data. So you’re asking a very prudent question more about implementation science. Our users in the U.K are giving us a lot of very helpful input. This is the first time they’ve had a diagnostic tool for burn of this quality and caliber. They’ve historically used laser Doppler imaging in the U.K., and they are comparing this to that tool. That technology has not taken off in the United States due to many limitations with that. Our device is designed to not have those limitations, and so it’ll be interesting how it compares in the U.S. But a lot of what we’re learning right now is how the device performs relative to their previous experiences.

We’ve had very positive feedback. I anticipate that there will be data shared at several of the upcoming scientific meetings as early as one in April from our colleagues in the U.K. As for guidelines, yes, there’s a lot of work done to establish coding for this product of this caliber. There’s going to be a lot of work that’s going to be needed to change care paradigms because we’ve never had a tool like this. You can imagine an orthopedic surgeon, you had a broken arm, you never had an x-ray, it’d be very different than once you got a next ray. So we anticipate that this is going to have a beautiful adoption, but that there’s going to be a need to really study the implementation and do it in a constructive, well-educated way.

Ramakanth Swayampakula: Thank you all. Thank you for taking my questions.

Michael DiMaio: Thank you, R.K.

Operator: This concludes our question-and-answer session. I would like to turn the conference back over to Dr. DiMaio for any closing remarks.

Michael DiMaio: Thank you very much. So thank you again for your participation and continued interest in Spectral AI. We are very pleased with the progress we’ve made and remain optimistic about our prospects for growth. We hope to speak with some of you in the future events, including upcoming meetings that are coming both locally and abroad, as well as the annual American Bird Association meeting. I really want to thank Dr. Carter and Dr. Wong for the time with us today as well. I thank everyone and I wish you a very good evening.

Operator: The conference is now concluded. Thank you for attending today’s presentation. You may now disconnect.