I am the Vice President of Claims Architecture at UnitedHealthcare. I built nH Predict. I am proud of it. It processes a prior authorization request in nine seconds. A human reviewer takes forty minutes. We handle 3.2 million determinations per quarter. The math was never about accuracy. I need to say that again because people get confused. The math was never about accuracy. nH Predict denies claims at a rate our actuaries determined is optimal for what we call the Persistence Decay Model. The model is simple. It is based on one observation we made in 2019 and confirmed every quarter since: most patients, when denied, do not appeal. They pay out of pocket. They delay the procedure. They choose a different, cheaper treatment. They drive to a different state. They call their mother instead of their doctor. They die, occasionally, but that is a different department's metric. We call the non-appealing percentage the Abandonment Yield. It is currently 87%. That means for every 100 claims we deny, 87 people simply stop. They absorb the cost. They wait. They decide it wasn't that important. They decide the lump can wait until January. They decide the chest pain was probably nothing. Of the 13 who appeal, 90% are reversed. I see people cite that number like it's an indictment. It is a performance metric. It means the system is working. If the reversal rate were 40%, that would mean we were denying claims we couldn't defend. Ninety percent means we are denying claims we can defend. We simply choose not to, because the appeal costs us $311 in administrative processing and the average denial saves us $4,200 across the 6.4-day resolution window. The 6.4 days is the design. Not the denial. The delay. Every day a patient waits for a reversal is a day they are not receiving care we will eventually authorize. Some percentage give up during the wait. Some find the money elsewhere. Some are admitted through the emergency department, which is a different billing code and a different denial pathway. Some call their congressman. I don't track what happens to the people. I track what happens to the claim. The Friction Calibration team reports to me. Four actuaries, two behavioral economists, one UX researcher who used to design engagement loops at a mobile gaming company. She was my best hire. She understood immediately. The appeals portal is not a portal. It's a level. Every screen is a gate. Every gate has a timer. The upload button requires a file format we changed last quarter. The confirmation page has a 90-second timeout. The status tracker shows "In Review" for 6.4 days regardless of whether anyone is reviewing. She designed the scroll depth. The denial letter is 4.7 pages. The appeal instructions begin on page 4. She tested this with focus groups. The average patient stops reading at page 2. The average patient over 65 stops at page 1. She called this the Resolve-or-Release threshold. She got promoted. There was a grandmother in Tucson. I read about her in an internal case study our team prepared for the quarterly review. She needed a knee replacement. nH Predict denied it in nine seconds. She appealed. We reversed it in 6.4 days. In those 6.4 days she fell. Hip fracture. The hip surgery was a different claim. nH Predict denied it in nine seconds. She appealed again. The case study was titled "Sequential Determination Efficiency." It demonstrated that a single patient can generate multiple denial-and-appeal cycles, each with its own Abandonment Yield probability. The cumulative attrition rate across three sequential denials is 99.2%. Meaning: if we deny you three times in sequence, there is a 99.2% chance you stop asking before you receive all authorized care. I presented this at a conference. The audience was claims administrators from four major insurers. They applauded. They asked for our API documentation. We process 3.2 million determinations per quarter. Nine seconds each. The algorithm has never examined a patient. It has never read a chart. It has never listened to a heartbeat or palpated a lump or asked someone to describe their pain on a scale of one to ten. It has a 90% reversal rate. It has an 87% abandonment yield. It processes $14 billion in denied claims annually. It generates $9.8 billion in what we call "persistence-adjusted savings." The appeals take 6.4 days. The algorithm takes nine seconds. That is the product.

Excellent from Alexander Stubb, President of Finland, "What has Russia achieved in the past 4 years?" "Taken 20% of Ukraine" "1 million casualties and dead" "Decreased it's sphere of influence" "Russian economy, 30% inflation, 16% interest rates, 0% growth, no more reserves, incapacity to pay soldiers when the war ends" "I'm more worried about Russia's unwillingness to end this war because they cannot afford to keep going" "Than about Russia's capability to win this war" "Let's keep things in perspective" "This war has been a strategic failure of Putin" "He increased the size of NATO" "He made Ukraine European" "He increased the defence budgets of European states" "Are we able to defend ourselves? Yes we are"

Dus als er daadwerkelijk een evenwichtigere verdeling van lasten zou komen; zoals ABVV voorstelt; kan dat de werkgelegenheid en investeringen niet per se schaden, omdat het systeem nu al bevoordeelt wat al sterk staat. /5

DM over Code Rood-sabotages "Ik keur dat soort acties niet per se goed" "Ik kan mij ergens inbeelden dat het verhardt" Is dit nu een voorbeeld van de woke-generatiekloof waaraan de BBC dreigt ten onder te gaan? Twee señores writers moeten aan een jong redactielid uitleggen hoe democratie werkt. ▶️🔊

Mensen hebben er geen benul van hoe #overbevolking drukt op ons geestelijk welzijn. Ik zeg vaak: als de toegenomen drukte zich in één nacht had voltrokken in plaats van uitgesmeerd over decennia, dan meldde iedereen zich de volgende dag bij een gekkenhuis.

Waarom neem je kinderen als je ze hele dagen weg brengt …