When a disease is rare, a positive test is usually a false alarm
The idea in one line
A test can be very accurate and still be wrong most of the times it says yes. It depends on how common the thing is, and almost everybody forgets to ask.
A man does the company screening package. A week later a number comes back marked positive. He is forty three, he feels fine, and he spends the drive home deciding how to tell his wife. Nobody at the lab told him the one thing he needed: that for a rare disease, a result like his is far more likely to be noise than news.
This is not a trick and it is not a small correction. It changes the answer from almost certainly ill to almost certainly fine.
The idea, and where it comes from
It is called the base rate. The base rate is simply how common something is before anyone tests you. The rule for updating a base rate with new evidence was written by Thomas Bayes, an English minister, and read to the Royal Society in 1763, two years after he died.
You do not need his algebra. You need one habit of mind: never judge a test result without first asking how common the thing is.
The question, and the arithmetic
In 1978 three researchers at Harvard Medical School asked doctors one question. Here it is, close to word for word.
A disease affects 1 in 1,000 people. The test wrongly says yes 5% of the time. Your result is positive. What is the chance you have the disease?
You know nothing else about the person. No symptoms, no history.
Do it with people instead of percentages. Imagine 1,000 people walk in and all get tested.
One of them has the disease. Assume the test never misses, so that one tests positive. Real cases found: one.
The other 999 are healthy. The test wrongly says yes to 5 percent of them, which is about 50 people. False alarms: fifty.
So 51 people leave the clinic with a positive result and only one of them is ill. Your chance of being the ill one is 1 in 51, which is about 2 percent.
chance it is real = real positives ÷ all positives
One divided by fifty one. That is the whole calculation.
Out of 1,000 people tested, who gets a positive result
How to read it. Longer bar is more people. Both bars are people holding a positive result in their hand, and they cannot tell each other apart. Note that this is generous to the test: it assumes the test never misses a real case. If the test also misses some, the real cases bar gets shorter and the picture gets worse, not better. Source: the arithmetic of the 1978 Casscells question, prevalence 1 in 1,000 and a false positive rate of 5 percent.
What the doctors answered
Sixty people at Harvard teaching hospitals were asked. Twenty medical students, twenty house officers, twenty senior doctors. Eleven of the sixty got it right, which is 18 percent. The single most common answer was 95 percent, given by 27 of the 60.
The correct answer was about 2 percent. Nearly half of them said 95.
You might hope that was 1978 and medicine has moved on. In 2014 a team at Harvard and Boston University asked the identical question to 61 doctors, trainees and students at a Boston hospital, and published the result in JAMA Internal Medicine.
Thirty five years later: 14 of 61 correct, 23 percent. The most common answer was still 95 percent, still given by 27 people. The middle answer across the group was 66 percent, which is 33 times the true value.
One senior cardiologist explained his answer by writing that the number does not depend on how common the disease is. That is exactly the thing it depends on.
Why it is true
Here is the engine of it. The test's error rate gets applied to an enormous crowd of healthy people, and the test's accuracy gets applied to a tiny group of sick people.
Five percent of 999 is fifty. One hundred percent of 1 is one. Fifty beats one. The size of the healthy crowd does all the work, and that crowd is invisible to you, because you only ever see your own result.
So the number people confuse is this. How often the test is right is not the same as how often a yes means yes. The first is about the test. The second is about the test and the world together. Doctors are not bad at arithmetic. They answer the question they can see and the crowd is not in the room.
Where it is weak, honestly
This does not mean tests are useless, and reading it that way would be a worse mistake than the one it fixes.
First, the base rate that matters is the base rate for people like you, not for the whole population. If you have symptoms, or a family history, or you are sixty rather than thirty, you are no longer 1 in 1,000. You might be 1 in 20. Then a positive result is serious and should be treated that way.
Second, a positive screening result is not a diagnosis. It is a reason to do a second and different test. And the second test starts from your new base rate of about 2 percent, not from 1 in 1,000, so it can settle the matter. The lesson is not to ignore the first test. It is to not act on it.
Third, the 1978 question is artificial. It assumes a perfect test and no symptoms, which is a laboratory, not a clinic. Real screening decisions also involve the harm of treating people who were never going to get sick, and those numbers differ by disease.
The part that contradicts the usual advice
Everyone says test early, test often, catch it before it grows. For a rare disease in a person with no symptoms, that advice mostly manufactures frightened healthy people.
Fifty of those fifty one people will go on to have scans, biopsies and months of worry for a thing they never had. Those follow up procedures carry real risk and real cost. More testing is not more information when the base rate is low. It is more noise, and noise has a price.
Where it shows up in your life
Your money. Someone tells you a signal has been right 90 percent of the time. Ask the second question: how often does the thing it predicts actually happen? A warning that is 90 percent accurate about an event that occurs 2 percent of the time will be wrong most times it fires. The same logic explains why your bank blocks your card on a normal purchase.
Your children. A teacher, or a short questionnaire, says your son may have a concentration problem. Take it seriously and do not accept it as settled. If the condition affects a few children in a hundred, and the screening sheet flags one in six, then most flagged children do not have it. Ask for a proper assessment, which is the second and different test.
Your health. When a result frightens you, ask the doctor two questions before anything else. Out of a hundred people like me, how many have this? And out of a hundred healthy people, how many does this test wrongly flag? A good doctor will welcome both.
Three things to try
The next time any number frightens you, ask how common the thing is before you react. Out of a hundred people like me, how many?
Treat a first positive result as a reason to look, never as a reason to act. Get the second test before you make one decision.
When anyone says a method is accurate, ask how often the thing it detects happens at all. Accuracy without a base rate is a half sentence.
Go deeper
Casscells, Schoenberger and Graboys, "Interpretation by Physicians of Clinical Laboratory Results", New England Journal of Medicine, volume 299, number 18, 1978, pages 999 to 1001. The original question and the 60 answers.
Manrai, Bhatia, Strymish, Kohane and Jain, "Medicine's Uncomfortable Relationship With Math: Calculating Positive Predictive Value", JAMA Internal Medicine, volume 174, number 6, 2014, pages 991 to 993. The replication 35 years later. Two pages, free to read, and the most convincing thing on this page.
Gigerenzer and Hoffrage, "How to Improve Bayesian Reasoning Without Instruction: Frequency Formats", Psychological Review, volume 102, 1995. Where the trick of counting people instead of percentages comes from.
Thomas Bayes, "An Essay Towards Solving a Problem in the Doctrine of Chances", read to the Royal Society in 1763. The origin, and harder going than anything above.
An American rate rise on Wednesday is now the biggest thing on your calendar
The market has flipped from expecting an American rate cut to expecting a rate rise this week
Oil above 100 dollars a barrel and hot wholesale prices pushed the odds of a United States rate increase at the 15 and 16 September meeting to roughly 70 to 90 percent, according to reporting on the futures market. American consumer prices then came in at 3.4 percent for August, unchanged from July, which did nothing to argue against a rise.
What it means for youThink through the chain rather than the headline. If American rates go up, holding dollars pays more, money leaves emerging markets, and the Central Bank of Egypt finds it much harder to cut. That is the opposite of what everyone here has spent the summer expecting. For your own money it means the 19 percent deposit rate is unlikely to fall in the next few weeks, so locking money in is less urgent than it looked. For the business it means one thing above all: stop building any plan that assumes cheaper money or a calmer pound this quarter. Price and payment terms should survive a worse currency, not a better one.
What Egyptian money earned, three verified numbers
How to read it. Longer bar is a bigger number, nothing more. Read this as a rough ranking, not a fair race, because the three periods are not the same. The EGX30 figure is this calendar year to Thursday's close, about eight and a half months. The deposit rate is an annual rate. Inflation is the rise over the twelve months to August. Two more cautions. The index is not a portfolio you can buy, and a year when shares beat a bank account by fifteen points is not a rule about next year.
MNT-Halan has filed to list its Egyptian arm, and the detail is better than last week's version
MNT Tech Holding for Financial Investments filed with the exchange on 8 September to list its full issued capital of 160 million pounds, which is 1.6 billion shares at 10 piastres par, on the main market. An offering is targeted for the fourth quarter of 2026 at a reported valuation of 900 million to 1 billion dollars for the Egyptian business only, with Citi and EFG Hermes mandated. The exchange publishes the notice for five working days before its listing committee reviews it.
What it means for youLast issue carried this from a weaker source and told you to treat the valuation as reported. Now it is confirmed by EnterpriseAM and Zawya with the filing date and the banks, so you can use it. The timing argument is unchanged and sharper. A company that wants to price an offering in the fourth quarter needs revenue that does not bring new credit risk, and a bundled subscription is exactly that line on their income statement. Ask for a distribution message and a revenue share. Ask for nothing to be built. If they say the listing has frozen all new work, you have lost one email.
The stock market fell on Thursday and lost 18 billion pounds of value
EGX30 closed Thursday 10 September down 0.39 percent at 56,280 points, while the equal weighted EGX70 rose 0.32 percent to 21,403 and EGX100 rose 0.17 percent to 27,997. Total market value fell by 18.2 billion pounds to 4.408 trillion. The index remains up about 34.6 percent this year.
What it means for youNotice the shape, not the direction. The big company index fell while the smaller company indices rose, which is the opposite of last week and means the selling is concentrated in the large names foreigners hold. That pattern usually means outside money is leaving, and outside money leaving is what moves the pound. Watch it as a currency signal, not as a stock tip.
OPay is reported to have applied for an Egyptian digital banking licence
The Nigerian payments company is reported to have filed with the Central Bank of Egypt for a digital banking licence as it expands outside Nigeria. The application is described as still under review, with no approval granted, no licence class stated and no exact filing date given.
What it means for youTreat this as a lead, not a fact. TechCrier is not on your source list, the article gives no date, and it says reportedly. Do not put it in a deck and do not repeat the valuation or the licence type to anyone. What you can do is open a relationship now, while they are the ones who need friends in this market. A payments company entering Egypt has to launch with something that makes its wallet worth opening in month one, and a family health subscription at 10 pounds a month is cheap content for them. Being early to an entrant costs you an email. Being late to one costs you the slot.
Two things checked and deliberately not printed as facts. First, an Arabic aggregator claimed the Central Bank cut rates by 200 basis points last week, with National Bank of Egypt and Banque Misr trimming certificate rates. That contradicts the last verified central bank record and no openable article supports it, so it is unverified. If you hear it this week, ask for the decision page on the central bank's own site before you believe it. Second, Fawry, Paymob, valU and e-finance produced no company news dated 10 to 13 September, so the window was genuinely quiet for them rather than under reported here. No global banking story could be pinned to an article inside the window either.
Health · 3.5 min
Public insurance added 62 facilities in one governorate while the ministry started firing people
New UHI facilities, Minya
62
announced 9 Sep
UHI contracted providers
581
earlier 2026, soft date
Drug repricing filings
170+
reported 1 Sep
Of those, reviewed
~150
reported 1 Sep
Possible price falls
30 to 40%
industry estimate
HealthTag points of care
4,000+
your own figure
Read the top two cells together and the gap should bother you. Sixty two facilities in Minya alone, against 581 contracted nationwide on a figure whose date could not be pinned down. Either the national number is badly out of date or the rollout has accelerated sharply. Both readings matter to you and neither can be settled from this week's reporting, so it is printed as a question rather than smoothed into a trend.
Sixty two health facilities entered service under universal health insurance in Minya
The Healthcare Authority announced 62 facilities entering service under the universal health insurance system in Minya, alongside an accreditation and digitisation programme. Contracted providers nationwide had reached 581 on an earlier 2026 count.
What it means for youThis is the story you should be tracking most closely and it is the one nobody at HealthTag talks about. Universal insurance is not a partner, it is a substitute. In every governorate where it lands properly, your core pitch of a discount on private care competes with something free at the point of use. But look at where it is landing. Minya, not Cairo. The rollout goes governorate by governorate and it takes years, which means your real map is the list of places it has not reached yet. Ask one question this week: which governorates are next, and is your provider network strongest there or weakest there? If your network is thin exactly where insurance has not arrived, that is the gap worth closing, not another fintech logo.
The deputy health minister made surprise visits in Qalyubia and referred a hospital director for investigation
Deputy Minister Amr Kandil carried out unannounced inspections of facilities in Qalyubia and referred a hospital director for investigation over absenteeism. The visits are part of a continuing service quality enforcement push.
What it means for youThe state is now measuring providers on whether staff actually show up. That is your opening with private providers, and it is a better opening than a discount. You already know, from prepaid bookings on labs, radiology and clinics, which of your partners honour appointments and which do not. Nobody else in the market has that. Sell it back to the good providers as proof, and use it to renegotiate with the bad ones. Pilot it with ten clinics, not a hundred.
The drug pricing overhaul is finished on paper and still waiting for the Official Gazette
The Egyptian Drug Authority's new mechanism replaces fifteen year old invoice rules with cost based pricing plus an investment margin. Prices do not move while the dollar stays inside a 10 percent band, sustained moves lasting up to six months can trigger a review, and locally made medicines are priced at 50 to 60 percent of the imported equivalent. The pharmaceutical division head says some prices could fall 30 to 40 percent, and localisation investment is put above 500 million dollars. More than 170 company repricing applications have been received and roughly 150 reviewed.
What it means for youLast issue told you this was coming. The new information is that it is nearly processed, not nearly proposed, and that a dollar band is built into it. Two consequences. If pharmacy prices fall 30 to 40 percent by regulation, your percentage discount on medicine is worth fewer real pounds and a member cannot tell your saving from the government's. That is the weakest part of your value story and it is about to get weaker. Second, the dollar band is a warning shaped like a rule: if the pound keeps sliding past the band, medicine prices go up under the same mechanism, and your cash categories are where you cannot see any of it happen.
The ministry opened a two day appeals window for rejected nursing school applicants
The Ministry of Health opened an appeals link for candidates rejected from fee paying technical secondary nursing schools, running Saturday and Sunday, 12 and 13 September. Nursing capacity is one of the binding limits on how fast universal insurance can expand.
What it means for youSmall item, real signal. The constraint on Egyptian healthcare is not buildings, it is staff. That is why appointments slip and why your members wait. It also caps how fast public insurance can take your market, which buys you time. Do not build anything on this. Just stop assuming the bottleneck is demand.
Two hospital projects got ministerial attention in the window
The health minister issued directives on the Riyadh Hospital project in Kafr El-Sheikh and a completion date was disclosed, and he chaired a follow up meeting on the redevelopment of the Coptic Hospital in Cairo. Bed counts, costs and timelines were not confirmed in either case.
What it means for youBoth are public capacity, not partnership targets, and the numbers are not published, so there is nothing to size. Printed only so you know the window was covered.
No chart on this page, on purpose. The available figures are 62 facilities, 581 providers, 170 filings and a 30 to 40 percent estimate. Those are four different units from three different dates, and any chart drawn from them would look like a finding and would be decoration. The six cells above carry the same numbers with their dates attached, which is the honest version. Also checked and not found: no global healthcare or digital health story dated 10 to 13 September could be tied to an exact article, only rolling tracker pages, so the global half of this page is absent rather than padded. And a search thread about an Aswan gastroenteritis outbreak turned out to be from September 2024. It is not a current story and should not be repeated.
Economy · 2 min
Nothing new was published, and that is the story
Urban inflation
14.5%
August 2026
Headline inflation
12.7%
August 2026
USD/EGP, bank sell
51.37
quoted Sat 12 Sep
EGX30
56,280
Thu 10 Sep close
PMI
49.6
August 2026
Deposit rate
19.00%
last verified hold
Dates are the period each number refers to, not the publication date. Friday and Saturday are the Egyptian weekend, so banks and the exchange were closed on 11 and 12 September. The exchange figure is Thursday's close and the bank rate is Thursday's quote carried through the weekend. On the deposit rate, the 19.00 percent level is solid and the date is not. An August hold at the same level is reported but again could not be opened as an article, so the cell says last verified hold rather than naming a month it cannot prove. Brent crude stays out of this grid and sits on the Prices page with a warning. Net international reserves stay out for a third issue, with the end of August figure of 57.21 billion dollars unchanged and no September update.
No Egyptian macro release landed between Thursday and this morning
Inflation for August was published on 10 September and is already in the last issue. There was no monetary policy meeting in the window, no new purchasing managers index reading, and no reserves update. The exchange traded one session and the banks were closed for two days.
How to read it. Each point is the rise in urban consumer prices against the same month a year earlier. This chart is unchanged from the last issue because no new month has been published, and pretending otherwise would mean inventing September. What matters is what August could not know. It was measured before the pound crossed 51.30 and before oil went above 100 dollars, so it is the last clean reading of the old conditions. January to April come from secondary reporting, so treat the early shape as indicative and May to August as the reliable part.
S&P Global Egypt PMI
How to read it. Above 50 the private sector is growing, below 50 it is shrinking. The dashed line marks 50. August at 49.6 is unchanged for a third issue because the September reading is not out. It is still the closest to neutral in seven months, and the next print is the first one that will contain the oil shock, so it is worth waiting for. January to April stay blank because those releases could not be verified, and they are left as gaps rather than filled in.
Still no oil chart and no USD/EGP chart, for the same reason as the last six issues. Verified points come from different outlets on different days and stitching them into a line would invent a series. Here is the thing to actually watch this week, and it is not an Egyptian number. If the American central bank raises rates on Wednesday, every assumption about a cheaper pound and cheaper borrowing here gets pushed out by months. Egypt cannot cut into a rising dollar without paying for it in the currency. One month of better inflation does not change that, and a good August print in front of a bad September is a trap, not a turn.
Prices · 2 min
Silver fell hard, gold stalled, and oil touched 110 dollars
Last available one day move for each, percent
How to read it. Bars to the right went up, bars to the left went down. Read the dates in the labels, because these are not all the same day. The three world prices are Friday's move, since global markets traded on Friday. The Egyptian index is Thursday's close, because the exchange was shut Friday and Saturday for the weekend. Comparing a Friday move to a Thursday move is not a clean comparison, which is why each label carries its own day rather than one heading claiming they all happened together.
Price
Level
As of
Gold 21K, Egypt
EGP 6,320/g
Sat 12 Sep, one source
Gold 24K, Egypt
EGP 7,222/g
Sat 12 Sep, one source
Gold, world spot
$4,385.61/oz
Fri 11 Sep
Silver, world spot
$64.90/oz
Fri 11 Sep
Silver 999, Egypt
unverified
—
USD/EGP, bank sell
51.37
quoted Sat 12 Sep
EUR/EGP, bank sell
59.79
quoted Sat 12 Sep
SAR/EGP, bank sell
13.68
quoted Sat 12 Sep
Bitcoin
about $77,200
Fri 11 Sep
EGX30
56,280
Thu 10 Sep close
Brent crude
above $100
Fri 11 Sep, see note
Read the warnings, they are the useful part. Egyptian 21K gold is printed this issue after being unverified last issue, but it comes from one outlet outside your normal list, so treat both Egyptian gold figures as indicative rather than exact. Egyptian silver is unverified on purpose: two outlets reported 98.03 and 116.92 pounds a gram for the same Saturday, a gap of nearly 20 percent, and picking one would look precise and would not be. The three bank rates are quoted on Saturday with the banks closed, so they are Thursday's rates carried over, and they come from outlets outside your usual list. Bitcoin is a Friday level, not a settled close, and it fell about 6 percent on the week. On Brent, one outlet reports a settlement of 104.61 dollars after an intraday high near 110, and no second source confirms the settlement, so the table says above 100 dollars, which is the part two approved outlets support. Do not quote 104.61.
Why they moved
Oil went above 100 dollars for the first time since May on Gulf escalation
Brent rose past 100 dollars a barrel, reported as high as about 105 to 110, on fighting near the Strait of Hormuz and reported strikes on Saudi oil facilities. The jump immediately raised the market's expectation of a United States rate increase.
Silver fell 2.6 percent in one day while gold barely moved
Gold spot added 2.61 dollars, or 0.06 percent, to about 4,385.61 dollars an ounce on Friday, its third straight weekly loss at about 1 percent down on the week. Silver dropped 1.72 dollars, or 2.58 percent, to about 64.90 dollars, widening the gold to silver ratio to roughly 67.6.
American consumer prices held at 3.4 percent, which was enough to keep the rate rise on the table
United States consumer prices rose 3.4 percent over the year to August, unchanged from July, on figures released Friday. The reading left expectations of an increase at the 15 and 16 September meeting intact.
Silver and bitcoin took the damage because both live on borrowed optimism
Higher expected rates hit silver harder than gold because silver has heavy industrial demand and industry slows when money costs more. Bitcoin fell about 6 percent on the week to roughly 77,200 dollars on the same logic.
Egypt is cutting its fuel import bill while oil is on fire
Three separate moves in one week, all aimed at buying less foreign fuel
Egypt raised refinery utilisation to 80 percent to cut petroleum imports, approved continued oil and gas production from fields whose contracts had expired, and saw crude loadings at Sidi Kerir more than double amid Hormuz and Red Sea disruption, according to the tracking firm Kpler. Taken together these are supply side responses to an import bill that gets worse every week oil stays above 100 dollars.
What it means for youFollow the money to where it lands on you. Fuel was about 52 percent of Egypt's import bill in 2025 at roughly 21 billion dollars. When oil rises, Egypt needs more dollars, the pound weakens, and every imported thing in a clinic gets dearer. That reaches you twice. Providers will ask to reprice, and families will cut the parts of health spending they think are optional, which is dental, optics and check ups, exactly your cash categories. Two practical moves. Lock provider terms now in pounds, for as long a period as they will sign. And stop selling check ups as prudent, because prudent is the first thing a squeezed household drops.
Egypt set a target of 144 percent more solar and 400 percent more wind capacity by 2029
The target was published on 10 September, in the middle of the oil spike. The reporting gives the growth rates and the 2029 horizon.
What it means for youNothing this quarter, and it is better to say so than to invent a link. One thing is worth taking from it. A 2029 target is a three year answer to a problem arriving this month, which is a fair description of most national plans and of most of your own roadmaps. If your answer to a currency problem is dated 2029, you do not have an answer.
The American rate decision lands on Wednesday and the market expects a rise
Reporting on the futures market put the chance of a United States rate increase at the 15 and 16 September meeting at roughly 70 to 90 percent, driven by oil above 100 dollars and consumer prices holding at 3.4 percent. A rise would be a reversal of what markets expected a month ago.
What it means for youThis is the single most important item in the issue and it is not an Egyptian story. A higher American rate pulls money out of Egypt, pushes the pound down, and takes any Egyptian rate cut off the table. Every partner conversation this week happens in that weather, whether the person across the table knows it or not. Use it. A partner who is nervous about the currency is a partner who wants revenue that arrives in pounds every month with no capital spent, which is precisely what a bundled subscription is.
Three openings this week
Email MNT-Halan this week, before the listing notice clears. The filing is confirmed, the offering is targeted at the fourth quarter, and the window where they still want easy recurring revenue is now. One page. A distribution message and a revenue share. Nothing built.
Map your provider network against the universal insurance rollout, not against population. Sixty two facilities went live in Minya this week. Find out which governorates are next and check whether your network is strong there. Where free public care arrives, your discount pitch weakens, and you should know that before your members do.
Sign provider terms in pounds, for as long as anyone will agree. Oil above 100 dollars plus a likely American rate rise means the next repricing request is a matter of weeks. A three or six month fixed term costs you nothing today and is worth real money if the pound slides.
One move today
Take the number that worried you most this month and ask how common the thing actually is
Pick one alarming number from the last month. A test result, a churn figure, a warning somebody gave you with confidence. Write down two things before you do anything else. First, out of a hundred cases like this one, how many really turn out to be the bad thing? Second, how often does this particular signal shout when nothing is wrong? If you cannot answer either question, you do not yet know what the number means, and you should not have made a decision on it. Page one shows why: when the bad thing is rare, a warning is usually a false alarm, and the crowd of false alarms is invisible because you only ever see your own. Doing this once turns a frightening number into a question. Doing it every time is most of what good judgement is.
Checked and not printed. A story circulating about Egypt raising fuel prices is dated Tuesday 10 March 2026, not September, so it is six months old and should not be forwarded as news. A report of a Houthi capture of an island near Bab el-Mandeb and a precautionary Saudi pipeline shutdown could not be tied to a single article, and Reuters and Associated Press could not be reached, so neither is printed. A Lebanon item existed only inside a live blog rather than a standalone article, so it is left out under the same rule that keeps roundups off this page. The world half of this page is therefore two American items and the oil price, which is thin, and thin is the honest result of a weekend.
Before Cairo Wakes · Issue 007 · Sunday, 13 September 2026
Built for Helmy, Growth and Strategic Partnerships, HealthTag. Six pages, about eighteen minutes. Every number carries the date it refers to and a link to the exact article. Nothing is estimated. Where a value could not be verified it says unverified, and where a chart point is missing it is left as a gap.
Sources this issue: EnterpriseAM, Zawya, Masrawy, Ahram Online, Al Jazeera, CNBC, USAGOLD, TechCrier, New England Journal of Medicine, JAMA Internal Medicine, Psychological Review, Goodreads, YouTube.