There are lots of studies in rats, of course. 

They have the advantage that the question-being-asked was chosen in advance, and the conditions selected so that the observations made would be as relevant as possible to that question, while minimizing the involvement of any other factors that weren't directly part of the cluster-of-tightly-related-factors that the scientists were actually trying to measure. They also have the advantage of scale; thousands of rats, across multiple experiments, were sacrificed toward the cause of greater understanding of biological processes. 

They have the major disadvantage that rats are not humans - and, in fact, the differences between rats and humans are most extreme in the brain. 

(Not all of the animal experiments were on rats specifically. A small handful even involved primates - but the cost-benefit analysis is very different, for more intelligent mammals like monkeys, and so the vast majority of the recent data is in rats or rat-equivalent-intelligence mammal test subjects.) 

This doesn't mean that the expert Treatment Planners or the medical prediction markets have nothing to go on!

Sure, they have less to go on than they would for a more common medical event, one that might occur millions of times every year. Especially anything where the relevant sensors have existed for years or decades, where they can feed historical datasets containing tens or hundreds of millions of cases to their models, and - given how statistical distributions work - expect that any given patient will fall within rather than outside of the range within which the resulting model outputs useful rather than garbage predictions. 

In that type of situation, the Diagnostic prediction markets can take in sensor data on a patient and, usually, confidently predict the prognosis, and response to a range of different treatments, to within a fairly narrow range. 

This is not one of those cases. This is a case where the data is limited, and the patient is legitimately outside the distribution within which their models can be expected to produce vaguely sane outputs. This is legitimately really inconvenient! Especially given the stakes, everyone involved here would prefer if, instead, they were in a situation where they did have really comprehensive data and actually-thoroughly-tested predictive models! Instead, they are stuck cobbling together disparate datasets, each one non-analogous to this case in a different way, and combining them via models that have literally never been tested under realistic conditions. 

But the Diagnostic system of experts and prediction markets isn't trying to get a good grade on a test. They're not trying to find a defensible Right Answer to this question, and then defend it. 

They are trying to - well, a lot of the individual players here are optimizing for making money on the prediction markets, as per the design - but the overall system is aimed at maximizing the patient’s chance of survival-without-brain-damage. The entire goal of the Diagnostic predictions markets is to, on an ongoing basis as new information comes in, maintain a best-guess model of the underlying biological reality, and then simplify that down to something that can be conveyed to the Treatment Planners and the on-site oppers, so that better decisions can be made in real time, in order to achieve better-in-expectation outcomes. 

(Under this degree of uncertainty, the experts and the markets are far more likely to make mistakes - mistakes that would be theoretically avoidable in a hypothetical world where somehow they had 100x as much high-quality data on this particular medical problem. No one disagrees with this. No one is going to blame the decision-makers in this situation if they make their best judgment call, and get it wrong. If anything, the default expectation is that they’re going to get quite a lot of things wrong under conditions like these. There will be acerbic rants, later, but not blame. Anyway, debriefs on the better processes they should have had are not the priority NOW.)

There is a lot of guesswork, and frantic repurposing of maybe-possibly-relevant models from obscure research projects, and quietyelling on conference calls. These are conditions under which even the best and brightest of dath ilan are likely to make mistakes, and the medical prediction market system is aware of that. 

At the end of it, though, there are numbers up on the screens that Merrin can see, and the numbers mean something, even if they aren’t quite as reliable as what she might expect in a more standard, well-understood, boring case. 

Merrin is expecting some updates on the Diagnostic prediction markets! She is expecting them to be negative updates! 

But the shift in probability estimates for survival and for various sub-outcomes – without brain damage, with only minimal brain damage, with moderate or severe brain damage, etc – are going to be in response to the exact same input that she just spent five seconds mentally freaking out about.

If she expected the markets to land on something worse - or better - than her current vague-gestalt sense, then she would already be making that emotional update. In fact (and as things should be) she has no idea which direction the additional expert knowledge will push things in. 

Anyway, Merrin isn't actually looking at the screens at all, because she's busy setting everything! Honestly, her top priority is the extracorporeal membrane oxygenation circuit, but she should probably set up the internal mechanical cardiac support first, even though for right now the external compressions-vest seems to be managing a surprisingly reasonable blood pressure. 

- actually Merrin (while not pausing at all in her setup) wants to explicitly flag for the Treatment Planners that she is currently feeling weirdly confused and uncertain about the correct prioritization here!

Not really on that one specific question - she should set up the cardiac support before the ECMO, given that the patient does not currently have a spontaneous heartbeat - but on a broader level. She...feels like she's been noticing some observations that feel in conflict? 

Updated diagnostic predictions over outcomes over outcomes:

- Estimated likelihood that the patient will survive until they reach the hospital: down, but only a little, back to 93%. 

- At least minor brain damage: up again by 18% from previous, new odds are 85%.

- Moderate brain damage: up from previous by 40%, new odds are 62%. 

- Severe brain damage: also up by 40%, new odds are 38%. 

- Some sort of complication happening in transit that causes additional anoxic injury or sets off an uncontrolled reperfusion-injury spiral before they have the ability to do anything about that: 45%. Rises to a near-certainty if they have to resort to rewarming the patient aggressively. 

(Which is still an option on the table, if he's otherwise imminently dying - they might still be able to keep the damage to the 'moderate' range, and get an outcome where the patient is significantly disabled but can at least make a meaningfully-informed decision on what to do about that. If he doesn't recover enough to make even basic medical decisions for himself at all, they'll...deal with that when they come to it.)  

Merrin is, as usual, not the first person to point this out, and in fact she is pointing it out - if she's even pointing at the same thing - in a less useful way than many other, more verbally fluent experts advising on this situation. 

(If she were the first person to raise a concern, the non-specificity of her comment would be more of be an issue. But Merrin - who despite all her other advantages is operating at -1 SD thinkoomph from the population median - is approximately never the fastest at raise an issue. Her low verbal fluency, and related difficulties with explaining her reasoning in a legible way, are known and documented in her dossier.) 

In an emergency, it's generally not useful or justified to keep bothering the on-site opper and asking them to unpack their reasoning any further than whatever they provided on their own. They don't need additional distraction, and would obviously have already said everything that was obvious to them. 

Backchannel: Merrin can handle the distraction, and probably can and will unpack more detail on her observations if someone explicitly asks! It is costly for her, sure, but mostly she just has a psychological block around thinking to provide that detail at all, because she thinks of herself as stupid and so doesn't expect anyone to find her more detailed opinions and analyses useful.

Anyone who was previously operating on the assumption that Merrin was a normal dath ilani medtech, should update on the fact that Merrin is neurodivergent, and in particular, Merrin will sometimes only dedicate explicit reasoning to a problem if someone asks her for the answer. 

...That being said, it is still costly for her, so they should only ask her to verbally unpack her confusion for them if, in fact, this will be decision-relevant information. 

(There is, obviously, a monetary bid being placed in the most-relevant policy market, by Personnel and people who know Merrin, to convey their sense of how costly vs how worthwhile it is to interrupt Merrin right now and ask her to try to unpack her illegible intuitions.) 

- nobody is bidding for interrupting Merrin to hear more of her reasoning, at least not in this specific case. (There is quite a lot of debate going on behind the scenes, but nothing Merrin should be bothered with.)

Merrin's main reference screen will display (in Baseline medical-jargon shorthand): they have received and acknowledged her report, it matches other expert opinions and is being taken into account as semi-independent confirmation of said opinions, and they are updating and acting on it accordingly. Her report is potentially decision-relevant for determining the medium-term protocol, but for now - as in the next ten minutes - her protocol is unchanged. 

??????

Merrin honestly has no idea how they seem to have apparently decided that her incoherent verbal report was meaningful! What do they even think she meant?? She personally wasn't sure at the point when she was saying it! Honestly she still isn't sure!! 

...She's not even actually busy right now, at least not mentally? They told her the protocol and priorities were unchanged, and so her hands are now following the incredibly deeply engrained motions of doing some ultrasound-guided minor surgery to place the internal mechanical cardiac pump. This genuinely does not require almost any of her conscious attention; she had already put in dozens of hours of sims on it before she was recruited to Exception Handling at all. 

The patient continues to look objectively terrible, but significantly less terrible than one might expect, at her! 

Merrin has to pause the external chest compressions vest for this procedure, which makes it much more obvious that the residual electrical activity they were previously detecting in the patient's heart is, one, definitely real and not a signal-noise artifact, and two, definitely still being propagated across the heart muscle!

The electrical pacemaker systems within his heart seem to be pretty definitively settled on ventricular fibrillation as the thing they want to do right now. Which is not at all useful for actually pushing blood around, but it's still very obvious to Merrin. The 'coarse' (higher-voltage and wider variation) signal on the briefly un-interfered with ECG might have seemed implausible, likely just a sensor error, if not for the fact that Merrin, while performing the procedure, can literally feel his heart muscle wriggling, weakly yet determinedly, like a fistful of tired-but-desperate earthworms. 

....did she somehow hallucinate a totally wrong number for his core body temperature??? Because Merrin definitely thought the most recent value for that was under 23 C and she was not expecting this level of cardiac-muscle enthusiasm at that temperature! 

(She did not hallucinate the number! Though he's now up to 22.8 C.)

Huh. Interesting and - actually kind of impressive? Both the fact that his temperature increased (if only a little) without their active intervention, and the degree of functional metabolic activity implied by muscle contractions she can literally feel. 

She still sort of wishes he wouldn't, it's messing with her sensors which is rude.

It is! And it might normally make the mechanical pump less effective, but they do not actually seem to be having any trouble with this. They now have a direct pressure-waveform transducer just after the left carotid artery branches off the aorta - the right carotid is taken for internal O2-sat-clip placement - and the patient's blood pressure almost immediately shoots up to 95/40. 

Presumably because this is also increasing perfusion to his lungs, his O2 sats are now climbing. 

Merrin does not even need the urgent warning from Treatment Planning to dial the settings way down! While it's a good sign that they can attain basically-normal vital signs at all, it's not actually the goal at this point! 

Protocol calls for systolic blood pressure below 60 and goal mean arterial pressure of 40-45, goal O2 sat between 70-80%, and very tight parameters on blood pH and CO2 concentrations (once they get those to within normal range at all). This would be way harder to juggle with just ventilator settings, but fortunately Merrin will have much more direct control once ECMO is in position. 

...For right now, it looks like dropping the O2 concentration on the ventilator to like 30% is enough to stop the climb, and get his sats back down to the mid-70s. Unfortunately she doesn't have the newer and fancier version of the vasodilator-constrictor device for controlling brain bloodflow more directly, so the patient's blood pressure does spend an entire 15 seconds very out of range - actually spiking to above 100 systolic - until the change in settings takes effect. 

Getting that device in position seems like the next priority then! Fortunately Vellis is certed in it and can do it while Merrin takes care of the other next priority and inserts a large-bore catheter into the patient's femoral vein for hookup to the continuous hemodialysis setup. She has a super precise protocol to follow on trying to gradually decrease the patient's sodium, which is currently ABSURDLY high and still rising on each test, he's now at like 185 mEq/L which is more than 25% out of parameters from the upper end of normal range.

This is actually much less of a disaster than it would be if the patient were out of range by that degree on the lower end - a drop in sodium levels below 100 mEq/L, especially if sudden, can very quickly result in more water entering cells - including brain cells - causing seizures, coma, and often irreversible fatal brain damage. If this had been a freshwater drowning, the patient's chances would be significantly worse. 

High sodium has the opposite effect; at the level the patient is at now, it does have severe neurological effects, by "dehydrating" the patient's cells. Levels above 160 mEq/L are often fatal without prompt, high quality treatment. Even without the anoxic damage, a level this high is enough to sometimes cause seizures and coma. The tail-risk danger is of a subcortical or subarachnoid hemorrhage, from actively-shrinking brain tissue straining and rupturing blood vessels. 

Correcting high sodium is also fraught, because the brain is actually very good at adjusting to out-of-parameters sodium levels - but the compensation mechanisms don't immediately return to their previous equilibrium, especially if the patient's sodium has been elevated for a long time, and returning to a "normal" plasma sodium concentration at a rate that outstrips the re-adjustment can cause many of the same risks that actual objectively-low sodium levels do. 

In this case, it was probably an acute steep increase, unless this guy is just chronically dehydrated - he's in fact more dehydrated than the other patients, and the most coherent of them did report that their boat has been low on fresh water supplies since the storm really got into gear, and Kalorm has a tendency to make sure other people are fed and watered before him. There's no way his level was this high before the near-drowning incident, but he might have been outside normal parameters all day.

Anyway. They do want to get him under 160 mEq/L - and correct the dehydration and fluid balance fully - before initiating the rewarming protocol, because they super have not tested whether the set of proteins used for it even work with physiological parameters that far outside the usual zone.

So! Very careful continuous hemodialysis, during which Merrin needs to delicately control a gradual shift in these dozen different electrolytes and other blood solutes – and keep a super close eye on the brain-damage-byproducts sensor, EEG reading, and regular brain ultrasound and CT scan data, and adjust the dialysate solute concentrations and thus the filtration rate as necessary if they think damage is occurring. 

The policy markets think this protocol should let them decrease the patient's sodium to below 160 mEq/L over the next four and a half hours, but they're actually only 70% confident of that; if the patient tolerates changes worse than expected, they're a) going to accumulate some additional cell damage in the process of learning that, and b) be left having to run the neuroprotective rewarming protocol with a much higher risk of something going badly wrong. 

Also, while Merrin does have fifty-plus hours of sim time on managing the controls simultaneously with 3-4 other machines, it's still going to be a LOT to track. 

And setup is obnoxious; she has to connect the tubes and filter and temperature-control box, and outflow for the waste "urine" produced after filtration, and inputs for administering extra fluids directly to the patient's bloodstream either before or after the filter - and all of that needs to be secured very thoroughly against acceleration. 

It takes her long enough that Vellis is ready to hand her the controls on the patient's carotid blood pressure by the time she's ready to actually hook things up. This is now three - no, four counting the ventilator - different machines she has to interoperate to keep the patient within basic vital-sign parameters. 

His blood pressure does briefly plummet when the circuit starts pulling blood from his body and returning only plasma, but Merrin has a lot of room to go up on the mechanical cardiac pump settings - which of course results in a few minutes of correcting-for-overcorrections before she has basic vital signs staying within parameters without constantly making adjustments, and can move on to focusing on ECMO setup. 

Diagnostic models are guessing that Kalorm's heart and its electrical control system are especially irritable, not just because of the cold and the still-out-normal-range acidity of his blood, but directly because of the out-of-range sodium and other electrolyte levels. Correcting this may actually have a significant effect.

They're considering directly administering a paralytic drug to the heart muscle, just to keep its oxygen needs down, but that has nonzero risks, and they may actually get a spontaneous organized rhythm back at some point. Market is putting 8% odds this will happen within the next 4h with just electrolyte correction and no other intervention, rising to 13% if they use the internal defibrillator system on the mechanical pump once his sodium is below 160 mEq/L, and 17% that using the electrical pacing system along with the pump would get them enough of a response to be worth trying. 

They do want Merrin to administer a very short-acting muscle relaxant (not intravenously, but via diffusion directly onto the surface of the patient's heart; the mechanical pump system comes equipped with a port for that) and see if this calms down the desperate worms and gets them a bit less sensor interference (and of course lower oxygen needs for the heart muscle, which is at risk of taking some damage over the hours they'll need to use the mechanical pump.) 

Yeah okay done, and now she can get the ECMO set up, and with it the levers she needs to shove the stupid blood CO2 levels actually within parameters. Also his lactic acid levels are clearing, but started so high that the latest result is STILL, like, above 8. Frustrating!

The hemodialysis will help clear them faster but she can't actually do it as fast as she wants, it's not a fully independent lever from the sodium and other electrolyte control, which means they're going to have to keep administering bicarbonate - in the form of sodium bicarbonate - to correct the patient's pH. Of course, once she gets the ECMO circuit set up, the CO2 should be resolved much, much faster. 

It's taking a weirdly long time to correct the patient's CO2 the normal way via his lungs? Like, Merrin feels vaguely confused, it feels like it's a harder battle than the lactate. 

She flags that confusion for the Diagnostic experts and markets while she gets set up again. 

...Yes, this was noted already, Merrin's intuition matches their bio models. 

The obvious explanation is that more metabolic activity is happening than they expected, even with all the direct observations of things like cardiac muscle activity and the surprisingly not-totally-absent EEG activity. The patient is just...slightly more metabolically functional than would usually be predicted by a body temperature of 22 C. It may be just part of the same cluster of observations Merrin made earlier, hinting at the patient - who did not, after all, come in with a particularly thorough medical chart - having been healthier at baseline than their generic bio-models are taking into account. 

Combining this with various other now-incoming sensor data is interesting, and changes some of the tradeoffs here. Merrin should anticipate some updates to the stabilization protocol. 

Well, she'll finish setting up her life-support equipment while they think about it. 

The ECMO circuit connection requires another incision in the patient's chest, a mildly nervewracking ultrasound-guided probe adventure around the heart to the vena cava, and placement of two even larger-bore shunts - and, once they're in place and the circuit is running, a clip to pinch off the vena cava between them, to mostly prevent oxygenated and de-oxygenated blood mixing. That's normally fine, actually, but in this case they want a lower and carefully controlled O2 sat in the blood entering the patient's left-sided heart and proceeding to the systemic circulation, most importantly the brain. 

It's really stressful to carry out in a gently-rocking medicopter, but Merrin totally does not slip and poke a hole in the largest vein in the patient's body! 

The patient's heart is especially not a fan of the ten-second interlude when blood is leaving his body before reaching the heart, but before any of it is re-entering from the circuit. (They didn't try to start it circulating with transfusion blood on an organ-transport machine, like with a particular previous case; it's fiddly and tricky enough to pull that off in a well-equipped ICU on totally stationary land, and they're prioritizing speed.)

Venous return to his heart drops rapidly to almost nothing, the sensors on Merrin's cardiac pump are so confused, and ramming up the settings - which Merrin does pre-emptively - doesn't even especially help, because the next four heartbeats have barely any blood to squeeze out, and the real time blood-pressure waveform from the carotid artery dives off a cliff. 

...As predicted, though, as soon as the oxygenated blood starts re-entering at the other end of the circuit and filling the patient's heart between beats, the stroke volume rises almost instantly back to its previous baseline. The actual measured blood pressure is less instantly resolved - the waveform almost looks like 3-4 seconds of asystole, a long "diastolic" slide down and down, bottoming out at a reading of 16 mmHg before the next pulse-waveform bump. 

On a more fragile patient, this would be enough of a perturbation to the system that Merrin would probably be spending the next hour using ever single tool at her disposal to keep the patient's circulation going at all. (If this patient had been in similar predicted-condition to the guy she treated in Harkanam, they would absolutely have taken elaborate extra steps to prevent that temporary drop, because it might not have been one his body could recover from at all.) 

On this patient, though - as was predicted with >95% confidence by the Diagnostic markets - it only takes around ten mechanically-pumped heartbeats, with the pump settings maxed, to bump the curve up and up, and within twelve seconds of when his blood pressure nosedived, the systolic is back above 50. 

Which isn't the same thing as being back in a stable equilibrium! Merrin is definitely working pretty hard for the next five minutes, staying on top of O2 sats and whatever superheated anomaly is going on with his pH right now. 

Apparently they're getting away with it! Merrin is only actually outside parameters for blood pressure for that one twelve-second period, which never even dropped to zero perfusion pressure, and twelve seconds in an awake patient isn't even long enough to cause a loss of consciousness. 

This patient's already-stunned cells could have been more fragile, but apparently they aren't; the cellular-damage indicators on the sensor have been dropping gradually, as the waste byproducts accumulated during his cardiac arrest are flushed out, and minimal ongoing damage is occurring to generate more of them. There's not even a blip on that trend.