Hospital administrators spend their days reading documents that contain claims: a vendor brochure, a compliance memo, a committee summary. The skill that separates a useful reader from a nervous one is knowing how much weight a claim can bear. ThreatOps, the working routine that combines threat intelligence and threat hunting, is mostly an organised version of that skill, applied to warnings about cyber attackers. This explainer uses no jargon that is not defined on the spot, and it stays strictly on the information-systems side of the building.
The routine is catalogued in the Ops Atlas page devoted to ThreatOps, a long reference that is open about one point worth repeating: the name has no official definition, so different organisations use it for different things. Some vendors even use it as a product brand. What the Atlas does is describe the underlying work and fix a clear scope around it.
The core idea: evidence has grades
Imagine two reports on your desk. One comes from a partner you have trusted for years and describes a pattern you have seen elsewhere. The other is an anonymous post claiming a group is about to target hospitals. You would not treat them identically, and a ThreatOps desk formalises exactly that instinct. Each source receives a grade for its track record and each individual report receives a grade for how believable it seems, so that a trusted provider’s first-ever feed can be scored as dependable in general yet still unproven in this instance.
The practical effect is that decisions can be traced. When someone asks why a firewall rule was added, the answer is written down: source, grade, date, reasoning. That is the same discipline a careful reviewer applies when reading research and deciding how far it supports a conclusion, and the site’s own overview of why accreditation matters when assessing hospitals abroad reads as a good example of weighing sources in a different field.
Words with numbers behind them
Analysts are encouraged to avoid vague phrases like “possibly” or “might.” The Atlas points to a vocabulary used in NATO circles in which each word corresponds to a probability band. “Likely,” for example, stands for something between sixty and ninety percent. A second dial, the confidence level, says how solid the underlying evidence is. The Atlas gives one firm rule: use one dial or the other in a given sentence, never both. Stacking “likely” with “moderate confidence” in the same breath muddles the message, so the two are kept apart.
Why would an administrator care? Because it turns a meeting from a debate about adjectives into a conversation about odds. If the desk writes “attackers are likely to try remote-access logins this quarter,” everyone hears roughly the same number.
Testing your own assumptions
The Atlas recommends two structured habits borrowed from professional analysis. The first is a Key Assumptions Check: before finalising a judgement, write out the premises you are leaning on, then ask which of them would sink the conclusion if false. The second is called analysis of competing hypotheses, reserved for contested judgements, where you line up several explanations and look for the evidence that is hardest for each to survive.
Both are low-tech. A whiteboard and thirty minutes suffice. They matter because the most expensive mistakes in this field come from confident stories that nobody questioned.
Overlap is not proof
Here is a striking illustration from the Atlas. In a lab exercise it compared three well-known attacker groups, each described in the ATT&CK knowledge base, and counted the distinct techniques they use between them: 92. Only two techniques appear in all three, logging in with legitimate accounts and entering through externally exposed remote-access doors. In other words, three very different groups share a tiny common core, and that core is made of ordinary behaviours that your own staff also exhibit every day.
The lesson for readers is twofold. First, seeing one of those behaviours in your logs says little on its own. Second, naming a culprit from partial overlap is risky; the Atlas calls attribution through group data “not a forecast,” and lists attribution theatre as an anti-pattern. A cautious desk describes what happened and what is likely next, and leaves the naming to others.
Where the tidy theory meets a messy case
The Atlas cites the 2024 joint advisory about the Volt Typhoon campaign as a caution. Government agencies observed that ordinary indicators, such as known bad addresses, were of little help, and that the intruders had kept footholds in place for at least five years. The point is that an empty sweep for known indicators cannot support the statement “we are not affected.” Absence of a signature is not absence of a problem. The right response is to hunt for behaviour, the way a careful reviewer looks for patterns, not just for a particular label.
What a null result means
This is perhaps the most counterintuitive part. When a hunt finds nothing, the Atlas insists it is not a failure, provided the data was good enough that you would have seen the activity if it had occurred. That is called a confident negative, and it is a real finding: it narrows the possibilities. When the data was incomplete, the hunt ends with a ticket for better records. Either way the hunt leaves an artefact behind, and an outsider holding only the write-up ought to be able to run it again.
How to use this as a non-specialist
You do not need to run a hunt to benefit. Three habits will carry over into everyday administration.
- Ask of any security claim: who says so, how reliable have they been, and how sure are they about this particular item?
- Prefer statements with a stated probability over statements with adjectives.
- When a team announces that a threat is “not relevant to us,” ask what they checked and whether the check could actually have found it.
For a sense of how patient, individualised planning is described elsewhere on the site, there is a readable article on how treatment planning is tailored at cancer hospitals; the comparison is only about the habit of matching a plan to the specifics, not about any clinical content.
Summing up
ThreatOps rewards modesty. Grade what you read, state your odds plainly, challenge your assumptions on paper and treat a clean sweep as a limited statement. Administrators who learn this vocabulary can ask better questions of their security teams, and teams that answer in the same vocabulary make fewer expensive mistakes. This article is general information on security practice and carries no guidance on any clinical matter.
