Now that one of the pressing problems has been addressed, and the aftermath navigated, there are a few other pending problems - nothing immediately urgent, but definitely things that need to be sorted out before they start trying to rewarm the patient. 

Delaying is worth it to the extent it buys them more time to prepare - for cleverer solutions to be improvised, for better equipment to be flown in as they realize they need it, for the Diagnostic prediction markets to anticipate what might go wrong in response to a planned intervention, and the Treatment Planner and markets to come up with plans for preventing those complications or at least a repertoire of responses in advance. 

Delaying is costly to the extent that the patient's condition is worsening, albeit very slowly - and because if the Venture Capitalist somehow does come through on his timeline and they're ready to start the improvised treatment protocol in just over six hours, they don't at that point want to be delayed on 'the patient still needs to be placed on a heart-lung replacement machine.' 

The benefits of acting sooner – if they can be confident of carrying out an intervention safely, and have reached the point of diminishing marginal returns to further preparation in terms of increased chance of success and decreased risk of making anything worse – is that they can maybe slow that deterioration even more. 

But there are still costs, even if a procedure goes off perfectly. Invasive interventions bear a cumulative risk, not just one-time, and many have an anticipated 'lifespan'. For example, machines to replace lung function use a high-surface-area array of fine tubules formed of semi-permeable biosynthetic membranes floating in a hyper-oxygenated nutrient fluid, imperfectly imitating the lung capillaries that absorb oxygen from the alvaolae. Flow pattern are different, often including unpredictable turbulence that can damage red blood cells, and unlike with biological capillaries, the biosynthetic membrane is inert, incapable of constricting or dilating to manage the pressure inside the circuit, and the membrane doesn't regenerate itself. Older models, based on earlier manufacturing technology, also lacked the quality control to consistently match the smoothness of real epithelial membranes, and microscopic irregularities would eventually start accumulating clots; the circuit had a lifespan of around 24 hours if the patient was on anticoagulants. The latest generation is much better, and can last several days even without taking the risk - which they can't afford for this patient - of using systemic anticoagulant drugs. 

The problem is that this is at normal temperatures. Among the many, many other biochemical processes altered at lower temperatures, blood viscosity increases, with obvious challenges when it comes to forcing blood through a high-resistance circuit of narrow-diameter imitation capillaries which can't dilate to reduce internal pressure. They can reduce pressure by reducing total flow, but slower-moving blood is also more prone to clotting. The current best models on how long a filter circuit will last at 20 C have very wide error margins, but the lower end is 10-12 hours – and 'while in the early stages of their improvised experimental rewarming protocol' is among the worst times imaginable to have to swap out the entire circuit (and every filter change costs the patient somewhere between 100ml and, if they get really unlucky, 500 ml of equivalent blood volume.) The later they can leave it, the less clock time they'll be putting in with the system operating well outside its design specs. 

That's just for lung function; the full version of the heart-lung replacement machine obviously also includes a pump, or in modern models, two pumps with separate controls - drawing blood from the patient's returning venous circulation, actively pushing it across the filter, and then running it past a second pump. This gives the medical staff very fine, and partially independent, control of both the pressure and flow inside the "lung" circuit, and the systemic blood pressure; it's the closest they have to imitating the interlocking natural biological control system of the combined circulatory-respiratory system. But artificial pumps, even with the current top-of-the-line technology, have pockets of internal turbulence that, again, damage red blood cells and platelets. Current equipment has a setup to catch and strain out some of those ruptured cells before sending them through the "lung" circuit to deposit clots in the tiny tubules, and before dumping them back into the patient's body - but they can't entirely catch the chemical byproducts released, including potassium and free-floating hemoglobin, which are also not great for the patient's circulatory system. 

The pumps will probably still work at 20 C, especially if they can design a software control system that more cleverly adapts flow rate to blood viscosity; frantic research is being carried out both on that and on modified mechanical components. But the rate of platelet damage and hemolysis of red blood cells will be higher – and right at the time when they're also dealing with wild swings in blood chemistry and accumulated cell-damage byproducts as the patient's own metabolism responds to the rising temperature. 

The current best idea is to attach a plasmapheresis circuit after the "lung" circuit and second pump, to filter out the non-cellular liquid component of the blood and - via specialized chemical substrates still frantically being tested - to selectively remove the protein components they don't want to dump back into the patient. But coordinating flow is going to be a pain, and obviously removing the liquid component of the blood means leaving behind much more concentrated red blood cells - plasmapheresis would usually filter plasma from the patient's bloodstream directly, rather than from the final stages of an already-elaborate circuit.

Also, doing it this way means that the final piece of equipment will be both rather jury-rigged, and also not something that literally anyone in the world is trained on using. They could instead use a more standard heart-lung replacement setup with only the hypothermia-relevant mods, and place a standard-equipment plasmapheresis circuit separately, but that gives them much less fine control on the overall system behavior. 

These are just some of the constraints currently being balanced. 

There is now a new minor distraction, almost certainly not important enough to claim the attention of someone important, but: Merrin's mother is calling Personnel! Her daughter was supposed to be at a con and didn't show and isn't answering her cellular texter?? Also there are, like, so many helicopters converging on the hospital and Merrin's mother is kind of worried about whatever's going on right now??? 

She wants to confirm whether her daughter is okay, whether she's likely to be working the entire day and missing family dinner tonight, and if so whether Irris should make her a nice homecooked lunch and bring it to the hospital again? 

A rather unusual medical emergency has occurred and Merrin is the only person on staff who has all of the certs required to handle it.  Further details gated behind a secrecy oath, unfortunately, the sort where you've got to take a grade-one oath just in order to be told which grade it is.  Merrin is likely working the entire day, and... yeah, should probably be fed at some point.

Under the circumstances, they're going to want to have it be a commercial precut-for-easy-eating homecooked lunch that Merrin can eat one-handed.  Possibly even a commercial homecooked lunch blended to where Merrin can suck it out of a tube with her hands free.  Does Irris know anything about Merrin's food preferences that aren't on file?  It would potentially save asking Merrin at a time when Merrin shouldn't be distracted.

Oh goodness is it that kind of day! Merrin sure has a tendency to get herself into Situations, doesn't she. (Irris says this with pride and fondness). In terms of normal non-blended-to-suck-out-of-a-tube food, Merrin isn't picky and will usually eat whatever you give her (unless it's eggplant cooked in anything but this one (1) way Irris knows, but don't give her commercially-cooked eggplant.) She tends to forget to ask for food if she's busy, though. Here is a list of the foods she is especially enthusiastic about. 

Yeah, this is a good time for the hospital to be nice to Merrin.  Like, not that they're usually cruel to her, but, extra-niceness today.  They'll get a chef on that.

(Admin doesn't say which rank of chef they'll be putting on that.)

Awwww that's good to hear. Irris will try not to worry too much. 

(She is pretty curious about exactly what situation Merrin has managed to get herself into, this is definitely some kind of personal record for weirdness, but secrecy oaths are not to be messed around with and she doesn't actually need to know.) 

Merrin does tend to forget about the existence of food when she's this distracted, and is also definitely not going to fully appreciate the quality of the chef, but the combination of 'delicious' and 'low effort to eat' is definitely going to help with actually getting the requisite calorie intake. 

Six hours in, they manage to bring online a higher-grade prediction market system. Merrin now has access to more predictions, faster, but not ones she can trust to the same extent as a real, tested, open market system. 

...One of the experts involved in setting it up starts giving her the quick explanation of the limitations, in terms they probably think are extremely simplified and accessible to non-specialists, and she sort of snaps at then that she has literally no idea what they're saying and less than zero interest in being taught new math right now. This would, under normal circumstances, be really embarrassing. Merrin has been throwing her brain at making sense of stupid data visualizations on various cross-sections of her sensor data logs during every single minute of not-entirely-occupied-by-reacting-to-problems "downtime" she can scrounge; she is sort of past caring who does or doesn't think she's stupid. It's not like it isn't already obvious, when she keeps repeatedly asking the data-visualization experts to redo something simpler because she's staring at it going cross-eyed and failing to even make sense of what it's depicting, let alone get the pattern into her subconscious to help guide her reactions.

(There's a certain mental state that Merrin gets into after enough hours of non-stop reacting. It's not exactly that she's tired, not in a reflex-slowing way, but it's as though some part of her gets emotionally depleted long before the point when she actually starts running low on stamina. The mental lever that produces embarrassment when whacked is gradually less and less responsive, and eventually - here, about six hours in - she reaches the point when she is what one might describe as "out of fucks to give."

It has some interesting effects. She's less inhibited, less polite, much more likely to respond with brief rudeness to being interrupted by a person. But she's also, in a way, less distracted. Her raw reaction times aren't faster, but the moments when she loses focus are rarer and briefer, and there's actually a small but measurable improvement in her performance. 

Tomorrow morning, she's going to wake up, remember all the times she was moderately rude to top world experts in obscure domains of medicine, and be mortified again, but she's hardly the only one occasionally engaged in some quietyelling in situations that, if they were under less prolonged stress, they would have been able to navigate a lot more gracefully.) 

Seven hours in, the Treatment Planners have a plan for transitioning to extracorporeal membrane oxygenation! Both for a system that works in the weird conditions they're throwing at it and will maintain an O2 saturation under 50% while running on a blood pressure of less than 50 systolic, and with 90% estimated likelihood that the the modified filter setup will hold up for at least 12 hours; the markets are currently at >95% that the rewarming treatment module and the team to run it will be ready by then. 

(Merrin did say out loud, two hours ago when they were discussing interim plans, that if they had to switch to high-frequency ventilation as a temporary measure, not only would it mean two incidents of having to re-juggle all her settings into a new equilibrium, she also has some sort of misophonia about it and it would be distracting for her. Apparently when you are the chief opper on a situation like this, your stupid preferences get taken seriously as an actual constraint? They had the equipment ready just in case, of course, but by very, very careful setting management, she's managed to eke out the last three hours without ever needing to bump the ventilator O2 concentration above 95%.) 

Merrin will be running the system once it's in place - and they're going to keep ventilating the patient's lungs – without the constraint of needing to maintain systemic oxygenation with that alone, so they can drop to a non-toxic O2 concentration and gentler volumes and per-minute respiratory rate, but it's not actually good for lungs to stop doing their job entirely, and it's not a sure thing that the patient's lungs are too badly damaged to recover once any of the normal cellular regeneration mechanisms are back online. 

Merrin will not be doing the actual setup procedure herself. She's certed for it, but it would take her full attention by itself, and it makes more sense to leave her on keeping everything else going while someone with exceptional fine motor skills and reflexes plus decades of hospital experience does the actual procedure. Also, they're modifying the protocol to make it possible that, up to the last minute, the opper can swap in a full heart-lung machine, if it turns out that the higher circuit resistance means that the mechanical cardiac pump can no longer maintain an adequate blood pressure.

(Merrin is not unusually advantaged at learning a new protocol on the spot, and is rather too busy to practice. At this point, she isn't even the person on-site with the most combined sim time on her current set of machines - just the one who has that plus eight hours of context on this patient specifically, and who can plausibly keep going for more than another 2-3 hours.) 

Also, rather than priming the circuit the normal way and hooking it up directly, they're going to prime it and then start circulation not on the patient, but instead through a module improvised from the organ-transport circulation support systems used for organ transplants. The transition will be awkward, with much narrower tolerances around timing, but it'll mean that the system is already flowing, already full of compatible blood oxygenated to exactly the desired degree, and the overall impact on the patient's blood volume, chemistry, and oxygenation status at the moment the connection is attached should be very close to zero. And it means they can connect the returning circulation a moment before the outgoing circulation, the opposite of how it's normally done, but predicted with 80% odds to provide a smoother and lower-risk transition in this particular case. 

There are other fallbacks in place if various different things go wrong, and personnel on-site (though not in the room, it's crowded enough) and ready to leap into motion on ten seconds' notice if they need to adjust plans. 

Merrin is still super on edge for this, but on the bright side, she has a protocol to follow in advance, bumping up the patient's systemic blood pressure just before the switchover. There's a planned ventilator-setting adjustment at the same time, but by default someone else will be doing that; Merrin will only take over if they have to deviate from the timing in the protocol suggested by Treatment Planning and start improvising. 

....Probably because of all the planning invested in this, though, the procedure goes almost perfectly. 

Now Merrin has a new, more detailed sensor suite and a new set of console controls! 

Which, of course, means that her process for wrangling all the machines is going to have to change a bit; the overall cognitive load and challenge of the next thirty minutes or so is significantly higher than it's been on average for the last two hours, and while not tired enough that she can't do it, Merrin is emotionally drained enough to find it frustrating. 

On the bright side, the extracorporeal membrane oxygenation console is great. It's a standard piece of equipment, not one that only shows up in weird obscure sims, so she has at least a hundred hours of sim time with it, with or without wrangling other systems at the same time. There are a lot more input variables to control than just the basic ventilator settings, but it's very intuitive for her by now, and the degree of fine control it gives her makes her feel like some sort of epic wizard. 

(Also, the ECMO setup has a temperature-control setting! It's not as effective as the full heart-lung setup will be - it would be pretty challenging to stick to the rewarming protocol with just that - but for keeping the patient at 20 C, it's a lot better than only being able to adjust the mattress temperature.)

She's still going to be pretty tunnel-visioned for the first thirty minutes or so of getting used to the new workflow. 

It's noticeable to the observers that this is pushing her closer to her limits; she's missing more alarms, defaulting to less sophisticated responses, and she's mostly not even trying to track the bigger picture; she'll need to be interrupted by a human if something changes in the treatment recommendations or the sensor data not directly linked to her machines, but that still calls for a shift in her overall strategy. 

But her raw reaction times are still maintaining. And while the Keeper shadowing her was getting to the point of probably outperforming her on her previous workflow, this is a new workflow, and Merrin still has twenty times as much emergency-conditions sim time on the ECMO controls. 

(There was some discussion of at least spotting Merrin for a thirty-minute break so she could rest before taking over again for the switchover, but Personnel was concerned that if Merrin knew her shadow was a rank-two Keeper, she - could probably still be convinced that she was the most qualified person for the next part, but it would at the very least be incredibly distracting and confidence-shaking for her. It was pointed out that they could just not tell her, but there are other considerations; on the recommendations of the prediction markets now focused on Merrin, and particularly on the betting of one particular expert in psychological modeling called Khemeth, having too much time to stop and think might by itself throw off Merrin's confidence. Whatever mental state she's in right now is clearly working, but may not be that robust to any poking. Also, it's not obvious that she needs the rest badly enough to make up for the context-loss of having to pick up after missing the last thirty minutes of the evolving situation; the Keeper will be leaning heavily on Keeper-specific training to manage stepping in with no ramp-up time.) 

Fortunately, by the time half an hour has passed, Merrin mostly has the hang of the new equilibrium. And still feels like an EPIC WIZARD. It's good for her mood. Probably so is the caffeine she requested. 

Only Merrin could be almost eight hours into a shift - which has been an emergency that could at any minute spiral out of control basically the entire time - and still look like she's HAVING FUN. 

(Merrin is mostly only having fun to the extent that she's pretending really really hard that this is a sim. She feels weird about having fun when it's a real patient.) 

8 hours in. Markets are now at 20%/50%/80% that the treatment module will be ready in 3.5/4.5/5 hours. How's Merrin doing on fatigue? 

Keeping her on this for another 5h will bring the total to thirteen hours which is insane to ask of a single person. But if they can be sufficiently confident that she won't burn out in that time, the rank-two Keeper can shadow her for a while longer until she feels more confident attention-splitting with the more complicated control suite on the ECMO, run a few times through a sim programmed with their best model of how the transition will look once the treatment module and team arrives, and then rest long enough to be in, if not peak, at least pretty adequate condition to actually take over whenever that happens.

(In the meantime, they can be ready to swap in one of the waiting replacements from another hospital, someone with equivalent sim time on the relevant machines. They can even make sure that there's always someone on standby, who's been watching Merrin's sensor suite and consoles long enough to sub in with some context. But not eight hours of context - and if they have to do that now, they'll be committing to doing it again in 2-3 h, and quite possibly again after that before the treatment module is actually ready.) 

Uh, honestly if things keep going this smoothly until the treatment starts, and they don't keep needing her to keep everything together while they do new and exciting things to the patient, she'll be fine? She could probably do another six hours; she's never done 14 hours straight, but she has done six-hour shifts followed by eight hours of sim time and not much of a break in the middle. She's never done sixteen hours in the same day and does not want to promise she can. 

(Emergency sims are way meaner than this, but they're often adversarial almost by design - they're meant to cram as much relevant troubleshooting practice into a 2-3h block as possible. Sims basically never give you six hours of everything being approximately under control, but this isn't a sim, and real life isn't actively out to get you in the same way.) 

Can she get some updated probability estimates on various complications that could happen before they start rewarming?

- Uhhh, she means can someone go interpret the markets for her and then explain it to her like she's literally an eight-year-old or something. She's not tired enough that her abstract reasoning is failing on that level, or anything, it's just that if it might be five more hours of this, she really wants to minimize cognitive load and avoid doing things that are hard for her.  

The highest-estimated-probability complication is that neural-cell-damage indicators will start rising at some point before the treatment module is ready - the markets are giving it less than 10% odds if treatment is somehow ready in two hours, but up to 30% odds if it takes more than five hours. If that happens, they have a flowchart of parameter changes and then various other measures to try; Merrin can review it, here it is, and of course someone will prompt her if this comes up. 

Nothing else has a >5% prediction on sudden-onset complications. The highest estimate for complications period is a GI bleed - 12% that this would happen within the next 5 hours, up to 35% if they somehow need to keep this up for another 8 hours - but they would almost certainly catch it before significant blood loss could occur, and there are responses planned. Emergency blood transfusions will likely mean that Merrin has to do a lot of rebalancing to stay within parameters. 

Uhhhhh Merrin cannot promise she has more than one more block of frantic troubleshooting left in her today? She's pretty sure she can handle it once, even if it's in four and a half hours, and keep up until they get back to something vaguely stable - but if it happens in ten minutes, she may actually be tired enough afterward for her reflexes to start going. 

...They can probably work with that. Probability estimates on an emergency troubleshooting level response being required from Merrin in the next hour are very low. 

(At this point it seems worth it to have the Keeper run sims and be fatigued for a few hours; if they do have an episode of patient deteriorating to resolve before the treatment module is ready - and Merrin is even right that her stamina will run out after that, the Merrin-related prediction markets seem to have more faith in her than she does - they'll tank the efficiency loss of subbing in a non-Keeper nurse for the remaining time.) 

Sure. 

Can she get a caffeinated drink please. And maybe a cookie?

(What Merrin actually kind of wants right now is a hug and for someone to tell her explicitly that she's doing a good job and not being an idiot and nobody is judging her for not being any good at prediction-market math, but she isn't about to ask for that out loud.) 

(There is in fact a prediction market on what will happen if various possible personnel go in and give Merrin a hug, and it's mostly predicted to be beneficial, but the expected benefit is small and the downside risk is large.)