Biomedical Information Technology, Software Development, and Informatics Support (BITSDIS 2)

Explore the BITSDIS 2 opportunity from the Department of the Interior, a $250M IDIQ focused on biomedical IT, software development, and informatics support. This episode breaks down the scope, competition landscape, and what vendors need to position themselves effectively for this high-value, full-and-open contract.

Listen now to stay ahead in federal contracting tune in now and start preparing your strategy before the April 30 deadline.

Contact ProposalHelper at sales@proposalhelper.com to find similar opportunities and help you build a realistic and winning pipeline. 

Transcript

A Cancer Cure Without Lab Coats

SPEAKER_00 0:00

Welcome to today's deep dive. So um imagine you're looking at a master blueprint to cure cancer.

SPEAKER_01 0:05

Aaron Ross Powell Right, but but this blueprint wasn't written by biologists in white lab coats, you know? And it doesn't contain a single chemical formula.

SPEAKER_00 0:12

Aaron Powell Not one. It was actually written by cloud architects, data engineers, and like artificial intelligence specialists.

SPEAKER_01 0:19

Yeah, it's a totally different approach. And today we are opening up some extremely dense, seemingly dry government procurement documents to show you how this works.

SPEAKER_00 0:29

Specifically, we're looking at a 2026 request for information, an RFI, and a draft statement of work. This is for a massive National Institutes of Health and National Cancer Institute contract. Trevor Burrus, Jr.

SPEAKER_01 0:41

Yeah. And it goes by the uh very bureaucratic acronym BIT DIS2. Trevor Burrus, Jr.

SPEAKER_00 0:46

Right. Okay, let's unpack this because if you read between the lines of these dense compliance frameworks, what you are actually looking at is just a staggering technological vision.

SPEAKER_01 0:55

Aaron Powell It really is. I mean, we are talking about the U.S. government's operational playbook for eradicating cancer using massive data pipelines, cloud computing, and AI. Trevor Burrus, Jr.

SPEAKER_00 1:03

Yeah, I like to think of it like this. Reading this RFI, it's like looking at the architectural blueprints for a massive futuristic city. But you know, instead of buildings and roads, it's a metropolis of data and algorithms designed to eradicate cancer.

The NCI Scale Behind The RFI

SPEAKER_01 1:18

Aaron Powell That's a great way to put it. And to grasp the sheer magnitude of what this document is asking for, you have to look at the entity driving the project.

SPEAKER_00 1:26

Aaron Ross Powell Right, which is the Center for Biomedical Informatics and Information Technology. Or CBIT.

SPEAKER_01 1:30

Exactly. CBIIT. Think of them as the well, the central nervous system for the National Cancer Institute.

SPEAKER_00 1:37

Aaron Powell And the NCI is huge. It's not just some boutique research lab.

SPEAKER_01 1:40

Oh, not at all. It is an organization with a six billion dollar budget. They have over 3,500 employees and collaborators.

SPEAKER_00 1:48

Wow. Six billion.

SPEAKER_01 1:49

Yeah. And CBIoT has the unenviable job of managing a dizzying, almost incomprehensible array of clinical, genomic, and real-world data across that entire network.

SPEAKER_00 2:01

Aaron Powell Which uh brings us to the first major hurdle outlined in these documents. Yeah. If you want to use artificial intelligence to find a cure, you first have to build an infrastructure capable of feeding that AI.

SPEAKER_01 2:11

Right. Because we often think of biomedical research as just, you know, doctors looking through microscopes.

SPEAKER_00 2:16

Exactly. But reading this, you realize that modern oncology is essentially a massive information technology bottleneck.

SPEAKER_01 2:24

It really is. The NCI is pulling together elements from these massive, globally distributed efforts, like the Cancer Moonshot and the Precision Medicine Initiative.

SPEAKER_00 2:33

Aaron Powell So they're dealing with a huge ecosystem.

SPEAKER_01 2:36

Yeah, an ecosystem where a tissue sample taken in a clinic in, say, rural Ohio needs to be correlated with an immune response data set generated by a monitoring center in Maryland.

SPEAKER_00 2:48

And a huge piece of that puzzle within this contract is a group called the Precision Medicine Analysis and Coordination Center.

SPEAKER_01 2:55

The PMAC.

SPEAKER_00 2:56

Right, the PMACC. Okay. According to the blueprint, this center manages current and next generation clinical trials using this master protocol platform called Madsechbox.

SPEAKER_01 3:05

Yeah, MadSeageBox is fascinating.

SPEAKER_00 3:07

So imagine you are a patient, right? Your diagnostic screening gets sent through a network of laboratories. They call it MDNet. That data hits the Matsy Cheachbox platform, and the system automatically determines if you are biologically eligible for specific highly targeted experimental therapies.

Feeding AI With Clinical Pipelines

SPEAKER_01 3:25

Aaron Powell But but think about the mechanical operational bandwidth required to run a system like Matchie Teachbox on a national scale.

SPEAKER_00 3:32

It's gotta be massive.

SPEAKER_01 3:33

It is. The request for information explicitly asks potential contractors to prove they can handle a scope of work requiring the equivalent of 150 to 200 full-time employees.

SPEAKER_00 3:45

Aaron Powell Wait, 150 to 250 E's?

SPEAKER_01 3:48

Yeah.

SPEAKER_00 3:48

Okay, I have to pause and push back on that premise. Why on earth does the National Cancer Institute need a 200-person tech army just to maintain a contract? Shouldn't those resources be going toward medical scientists?

SPEAKER_01 4:02

Well, what's fascinating here is that you cannot have a physical biological cure without the digital infrastructure to map it out.

SPEAKER_00 4:08

Okay, explain that.

SPEAKER_01 4:09

Well, a great doctor cannot cure a patient if they cannot securely access, organize, and analyze the patient's genetic data.

SPEAKER_00 4:15

Aaron Powell I mean that makes sense.

SPEAKER_01 4:16

Yeah. The documents detail systems like the Cancer Immunologic Data Center and the Correlative Study Management System, or CIDC and CSMS.

SPEAKER_00 4:26

Right, lots of acronyms.

SPEAKER_01 4:27

So many acronyms. But these aren't just digital filing cabinets, they are dynamic tracking systems. They monitor physical biospecimens like actual pieces of human tissue sitting in freezers, and they marry them to petabytes of genomic data.

SPEAKER_00 4:42

Aaron Powell So the biology and the technology are completely interdependent at this point.

SPEAKER_01 4:45

Exactly. You can't have one without the other.

SPEAKER_00 4:47

Okay, so you hire the 200 IT workers, you build the massive server farms, and you start collecting millions of tumor samples from across the globe.

SPEAKER_01 4:55

Aaron Ross Powell Right.

FAIR Data And The Vocabulary War

SPEAKER_00 4:56

But the moment you turn the network on, the whole thing grinds to a halt. Why? Because the lab in Boston and the lab in California are speaking two entirely different languages.

SPEAKER_01 5:06

Yeah, the data semantics problem.

SPEAKER_00 5:07

Aaron Powell Exactly. If Boston codes a tissue sample as tumor A and California calls the exact same biological mass growth B, a centralized database simply cannot compute the connection.

SPEAKER_01 5:18

No, it just throws an error.

SPEAKER_00 5:19

Right. It sounds like trying to solve the tower babble, but for pancer research.

SPEAKER_01 5:23

And preventing that tower babble scenario requires ruthless systemic enforcement. You know, you can't just ask thousands of independent researchers nicely to use the same words.

SPEAKER_00 5:33

Aaron Powell Because they won't.

unknown 5:35

Right.

SPEAKER_01 5:35

So the contract mandates strict adherence to what are known as the FAIR principles.

SPEAKER_00 5:41

F-A-I-R.

SPEAKER_01 5:42

Yeah. That means every piece of data must be findable, accessible, interoperable, and reusable.

SPEAKER_00 5:49

Aaron Ross Powell But how does a massive federal contract actually force a researcher in a private university lab to comply with that? I mean, practically speaking.

SPEAKER_01 5:57

Aaron Powell By controlling the digital vocabulary.

SPEAKER_00 5:59

Okay.

SPEAKER_01 5:59

Task Area 8 focuses on semantic infrastructure. The contractor has to maintain the NCI thesaurus, which is a curated standardized vocabulary for cancer-related terms, and the NCI metathesaurus.

SPEAKER_00 6:11

So it's like an official dictionary they have to use.

SPEAKER_01 6:13

Basically, they implement this using common data elements or CDEs, which are stored in the Cancer Data Standards repository, the CETISR.

SPEAKER_00 6:21

Wow. So essentially the IT contractor builds the universal translator.

SPEAKER_01 6:25

Exactly. If a researcher wants to upload data into the NCI's ecosystem, their software must automatically map their local terms into these rigid common data elements.

SPEAKER_00 6:36

It's brilliant, really. And it goes beyond just standardizing the names of diseases, too. It is also about standardizing how the actual clinical trials are organized.

SPEAKER_01 6:45

Yeah, that's Taskarian 9.

SPEAKER_00 6:46

Right. The documents outline the clinical trial reporting program, that's CTRP. This tracks every single NCI supported trial.

SPEAKER_01 6:54

The CTRP program management office acts as the ultimate traffic cop for the research ecosystem.

SPEAKER_00 6:59

A traffic cop. I like that.

SPEAKER_01 7:01

Yeah. I mean, think about the waste if two brilliant labs unknowingly spend millions of dollars running the exact same clinical trial.

SPEAKER_00 7:08

Well, that would be terrible.

SPEAKER_01 7:09

Right. So this program management office ensures every trial is registered, the data is standardized at the point of entry, and the results are accurately reported. They're looking for research gaps to fund and duplicative studies to prevent.

SPEAKER_00 7:22

Makes total sense. But the sources also focus heavily on the real world data program, which is completely separate from pristine controlled clinical trials.

SPEAKER_01 7:31

Right, the RWD program.

SPEAKER_00 7:33

It pulls information directly from electronic health records, EHRs, and regional cancer registries. They are scooping up the messy, everyday realities of patient care.

SPEAKER_01 7:44

Which is exactly why the semantic standardization we just discussed is the linchpin of the whole operation. Trevor Burrus, Jr.

SPEAKER_00 7:49

Because real-world data is so messy.

SPEAKER_01 7:52

It's chaotic. A doctor's hastily typed note in an electronic health record might have misspellings, abbreviations, or shorthand. Trevor Burrus, Jr.

SPEAKER_00 8:00

Right. They're rushing between patients.

SPEAKER_01 8:01

Exactly. So without that metathesaurus to translate an everyday doctor's note into a structured machine readable data point, that real-world evidence would be completely invisible to an oncologist searching the database for patterns.

Trial Tracking And Real World Evidence

SPEAKER_00 8:14

Invisible. Wow. Okay, so let's step back and look at the machine we've built so far. We have the massive cloud infrastructure, the petabytes of data flowing in globally, and we've forced all of it to speak the exact same standardized language. Aaron Powell Right.

SPEAKER_01 8:27

The foundation is laid.

SPEAKER_00 8:28

Yeah. Now we unleash the algorithms. This brings us to the most futuristic part of the blueprint. The NCI is acquiring cutting-edge tools to aggressively analyze this data matrix.

SPEAKER_01 8:41

Specifically in Task Area 13.

SPEAKER_00 8:43

Right. The contract heavily emphasizes artificial intelligence and machine learning.

SPEAKER_01 8:47

Trevor Burrus, And the applications the government is asking these contractors to build aren't theoretical. They are highly applied. For example, automated cancer detection using radiomic imaging.

SPEAKER_00 8:57

Okay, let's drill into that. We aren't just talking about a computer organizing files here.

SPEAKER_01 9:01

No, not at all.

SPEAKER_00 9:01

Aaron Powell We are talking about AI models analyzing massive data sets of X-rays, MRIs, and CT scans at the pixel level, right? Searching for microstructures of tumors that a human radiologist's eye might naturally miss.

SPEAKER_01 9:15

Aaron Powell Precisely. And they are pairing that visual analysis with genomic variant classification.

SPEAKER_00 9:20

Aaron Powell How does that work?

SPEAKER_01 9:21

Aaron Powell Well, the AI looks at a patient's DNA sequence, identifies a specific genetic mutation, and classifies how that variant might respond to a highly targeted precision oncology drug.

SPEAKER_00 9:32

Aaron Powell That is incredible. But uh here's where it gets really interesting. Oh, yeah. The requirement that really made me do a double take was seeing large language models LLMs explicitly mandated in this federal procurement document.

SPEAKER_01 9:44

Aaron Powell Yeah, that surprised a lot of people. Trevor Burrus, Jr.

SPEAKER_00 9:46

Right. Like we use LLMs every day to write emails or summarize articles, but the NCI wants to use natural language processing to extract insights from unstructured clinical text.

SPEAKER_01 9:56

Aaron Powell Well, think about the mechanics of how an LLM processes text. It doesn't just use a like control F function to search for the keyword cancer.

SPEAKER_00 10:05

Right, it's smarter than that.

SPEAKER_01 10:06

Exactly. An LLM understands semantic relationships. If a doctor writes a messy paragraph stating a patient took an experimental drug on Tuesday and reported severe nausea the following Monday, a traditional database cannot link those two events.

SPEAKER_00 10:20

Aaron Powell Because they're just words in a paragraph.

SPEAKER_01 10:22

Right. But an LLM reads that unstructured paragraph, understands the chronological and causal relationship between the drug and the symptom, extracts those insights, and turns them into a structured data row that the NCI can analyze.

SPEAKER_00 10:36

That is huge. They also want AI-powered clinical trial matching to completely automate patient recruitment, taking the burden off the local doctors.

SPEAKER_01 10:45

Yeah, which is a massive bottleneck right now.

SPEAKER_00 10:47

But reading all this, looking at how deeply the AI is penetrating the medical records, a glaring question comes up.

SPEAKER_01 10:54

Security.

Radiomics Genomics And LLMs

SPEAKER_00 10:54

Yeah. We are talking about people's most private, sensitive medical histories. Is it actually safe to let a large language model loose on personally identifiable health information?

SPEAKER_01 11:06

It is the single biggest risk factor in the entire operation. And the RFI tackles it with severe, non-negotiable security constraints. The NCI mandates strict AI governance. Contractors have to build within trusted frameworks and adhere to the National Institute of Standards and Technology, the NIST AI risk management frameworks.

SPEAKER_00 11:27

Aaron Ross Powell Sure, but mechanically, how does that protect a patient? How do you train an incredibly powerful AI on a patient's tumor history without the AI inadvertently memorizing and leaking that patient's name and address?

SPEAKER_01 11:38

Aaron Powell Yeah. So the core mechanism the contract demands is the zero retention-based model.

SPEAKER_00 11:43

Zero retention.

SPEAKER_01 11:44

Yes. This is absolute baseline for handling any personally identifiable information or protected health information, PII and PHI.

SPEAKER_00 11:53

So it's essentially like an amnesiac doctor.

SPEAKER_01 11:56

I love that analogy. Yeah.

SPEAKER_00 11:57

The AI model reads your file, diagnoses your complex genomic mutation perfectly, extracts the scientific insight to help the broader research, and then immediately zeroes out its memory of the actual text it read.

SPEAKER_01 12:10

Exactly.

SPEAKER_00 12:11

The weights of the model retain no memory of who you are.

SPEAKER_01 12:13

Aaron Powell That is exactly how it functions. The AI derives the pattern but discards the personal anchor. Furthermore, they are demanding what is called explainable AI or X AI.

SPEAKER_00 12:23

Oh, I found that fascinating. Because if an AI model acts like a black box and simply spits out a recommendation saying, prescribe this patient drug X, a human doctor cannot ethically just say, well, the computer said so.

SPEAKER_01 12:35

Right, they'd lose their medical license.

SPEAKER_00 12:36

Exactly. So explainable AI means the machine has to show its math, right?

SPEAKER_01 12:40

Aaron Ross Powell Exactly. It has to trace its logical steps. It must show the doctor the specific genomic markers, the exact clinical trial data, and the specific historical patterns it used to arrive at that recommendation.

SPEAKER_00 12:53

Aaron Powell So it's building trust.

SPEAKER_01 12:55

It has to. The tool has to build trust with the human physician, and you cannot build trust with a black box when a patient's life is on the line.

SPEAKER_00 13:03

Aaron Powell Okay, so we have this incredible, almost utopian vision, a globally connected, AI-powered, standardized data matrix actively hunting for a cure.

SPEAKER_01 13:14

Sounds great.

SPEAKER_00 13:15

It does. But translating that grand vision into a functional day-to-day government IT project is a brutal reality check. Because if you've ever worked in corporate IT, you know how hard it is to migrate a single database without crashing the whole office.

SPEAKER_01 13:30

Oh, it's a nightmare.

SPEAKER_00 13:31

Imagine doing that for the National Cancer Institute.

SPEAKER_01 13:33

Aaron Powell Right. And this is where the aspirational science meets the rigid logistics of federal procurement.

Zero Retention And Explainable AI

SPEAKER_00 13:38

Yeah, specifically in Task Area 5 for development and task area 7 for implementation.

SPEAKER_01 13:42

Exactly. The NCI isn't just looking for brilliant AI coders, they need enterprise IT heavyweights. The implementation tasks require deep expertise in agile methodologies, DevOps practices, and massive cloud hosting across the big players.

SPEAKER_00 13:58

Trevor Burrus, Jr.: Amazon Web Services, Microsoft Azure, Google Cloud. All of them. And while they are building this shiny cloud-based future, they still have to keep the legacy systems running.

SPEAKER_01 14:09

Oh yeah, you can't forget the legacy systems.

SPEAKER_00 14:11

The documents explicitly state that contractors must maintain specific older platforms. For instance, there's a system called C3D, the Cancer Central Clinical Database, which runs on older Oracle clinical architecture.

SPEAKER_01 14:24

Right. And you can't just unplug that while you build the new AI system.

SPEAKER_00 14:27

No.

SPEAKER_01 14:27

If you unplug C3D, active clinical trials lose their data streams. The research goes blind. Wow. And wrapping around all of this development is an absolute fortress of federal security requirements. The acronyms are DENSE, FITARA, FISMA, and FedRAMP, but what they represent is continuous, relentless surveillance of the networks.

SPEAKER_00 14:46

Let's explain that FedRAMP mechanism because it's not just a one-time security stamp.

SPEAKER_01 14:50

No, it's ongoing.

SPEAKER_00 14:51

A contractor can't just rent server space on AWS like a Silicon Valley startup. They have to build a digital fortress around that cloud partition. They are tracking every single ping, every data transfer to maintain what the government calls an authorization to operate or an ATO. Trevor Burrus, Jr.

SPEAKER_01 15:08

Right. The ATO is everything.

SPEAKER_00 15:10

If they fail in audit and lose that ATO, they are legally disconnected from the network.

SPEAKER_01 15:14

Trevor Burrus, Jr. Immediately. And that leads to the most vulnerable moment in the entire contract lifecycle: the handoff.

SPEAKER_00 15:21

Aaron Ross Powell Oh, this part was wild.

SPEAKER_01 15:23

Yeah. The RFI heavily questions potential bidders on their knowledge transfer capabilities and demands a rigorous transition in and transition out plan.

SPEAKER_00 15:31

Aaron Powell Let's talk about the stakes of that handoff. Because the government is basically planning for the breakup before the marriage even begins.

SPEAKER_01 15:37

Aaron Powell, they have to.

Legacy Systems FedRAMP And ATO

SPEAKER_00 15:38

If a massive tech company wins this contract, runs it for five years, and then loses the next bidding war to a rival company, they have to hand over the keys to the entire data metropolis. If the outgoing contractor is bitter or disorganized and they mess up the transition, it doesn't just mean a software update gets delayed. It means a live clinical trial for a real patient waiting for a cancer drug goes dark.

SPEAKER_01 16:04

The IT logistics literally keep patients alive.

SPEAKER_00 16:07

It's mind-blowing.

SPEAKER_01 16:08

And if we connect this to the bigger picture, this is why the RFI puts so much emphasis on task area one for contract management and task area fifteen for transition out. Okay. The government views mitigating organizational conflicts of interest OCI and enforcing that knowledge transfer as just as critical as the AI coding itself.

SPEAKER_00 16:27

Because it is. They are wrestling chaotic global medical data into submission. They are forcing the world's researchers to speak a single standardized language of oncology.

SPEAKER_01 17:13

Through the FAIR principles and the CDEs.

SPEAKER_00 17:15

Right. They are deploying explainable AI and large language models to pull invisible patterns out of that data. And they are building a rigid, highly secure logistical framework to manage the army of tech workers required to keep the lights on.

SPEAKER_01 17:30

It proves a fundamental reality about the future of medicine. You know, whether you are a database administrator, a cloud security architect, or just someone hoping for a future where cancer is universally treatable, this document makes it undeniably clear. Yeah. The future of the cure relies just as much on data engineers and software architects as it does on the biologists and doctors in the clinic. They are two halves at the exact same vital mission.

Contractor Handoffs That Risk Trials

SPEAKER_00 17:52

Which leaves me with a final thought I want to pose to you, the listener, as we wrap up this deep dive. Something to mull over that builds on that fascinating AI section we talked about.

SPEAKER_01 18:02

Oh, this is a good one.

SPEAKER_00 18:03

Yeah. Hidden in a requirements, the contract asks contractors to build pipelines to generate synthetic clinical data.

SPEAKER_01 18:11

Right. Synthetic data.

SPEAKER_00 18:14

This means using AI to create highly realistic but entirely artificial patient health records. It is completely fake data that mathematically mimics the biology of a real human being.

SPEAKER_01 18:27

And that allows researchers to train algorithms without ever touching real patient files, right?

SPEAKER_00 18:31

Exactly. It gets around the privacy issue entirely.

SPEAKER_01 18:34

Yeah.

Synthetic Data And Virtual Trials

SPEAKER_00 18:34

But follow that trajectory into the future. If these synthetic data models become perfectly accurate at mimicking human biology, disease progression, and chemical reactions, could we eventually reach a point where the National Cancer Institute tests thousands of experimental treatments entirely inside a cloud simulation? Could we see a future where millions of virtual trials are run and a definitive cure is validated in a data metropolis before a single human patient ever has to endure taking an experimental pill?

SPEAKER_01 19:02

That is the ultimate promise hidden inside these blueprints.

SPEAKER_00 19:05

It really is. Thanks for taking the deep dive with us.