Watson bai ciji likitan ba, kuma sosai

Ko da yake, kamar yadda a cikin sauran fannoni da yawa, sha'awar maye gurbin likitoci tare da AI ya ɗan ragu kaɗan bayan jerin gazawar ganowa, har yanzu ana ci gaba da aiki kan haɓakar magungunan AI. Domin, duk da haka, har yanzu suna ba da babbar dama da dama don inganta ingantaccen aiki a yawancin yankunansa.

An sanar da IBM a cikin 2015, kuma a cikin 2016 ya sami damar yin amfani da bayanai daga manyan kamfanonin bayanan marasa lafiya guda hudu (1). Shahararren, godiya ga rahotannin kafofin watsa labaru da yawa, kuma a lokaci guda aikin da ya fi dacewa ta amfani da fasaha na wucin gadi daga IBM yana da alaka da oncology. Masana kimiyya sun yi ƙoƙari su yi amfani da ɗimbin albarkatun bayanai don sarrafa su don mayar da su cikin ingantattun hanyoyin magance cutar daji. Manufar dogon lokaci ita ce ta sa Watson alkalin wasa gwaji na asibiti da sakamako kamar yadda likita zai yi.

1. Ɗaya daga cikin abubuwan gani na tsarin kiwon lafiya na Watson

Duk da haka, ya zama cewa Watson ba zai iya komawa ga littattafan likitanci da kansa ba, kuma ba zai iya fitar da bayanai daga bayanan likitancin marasa lafiya na lantarki ba. Sai dai babban zargin da ake masa shi ne rashin iya kwatanta sabon majiyyaci da kyau tare da sauran tsofaffin marasa lafiya da kuma gano alamun da ba a iya gani a kallon farko.

Akwai, da gaske, wasu likitocin ciwon daji waɗanda suka yi iƙirarin cewa sun amince da hukuncinsa, kodayake galibi dangane da shawarwarin Watson don daidaitattun jiyya, ko kuma ƙarin, ƙarin ra'ayin likita. Mutane da yawa sun nuna cewa wannan tsarin zai zama babban ɗakin karatu mai sarrafa kansa ga likitoci.

A sakamakon ba sosai m reviews daga IBM matsaloli tare da siyar da tsarin Watson a cibiyoyin likitancin Amurka. Wakilan tallace-tallace na IBM sun yi nasarar sayar da shi ga wasu asibitoci a Indiya, Koriya ta Kudu, Thailand da sauran ƙasashe. A Indiya, likitoci () sun kimanta shawarwarin Watson don lokuta 638 na ciwon nono. Matsakaicin yarda don shawarwarin jiyya shine 73%. Mafi muni Watson ya fice a Cibiyar Kiwon Lafiya ta Gachon da ke Koriya ta Kudu, inda mafi kyawun shawarwarinsa ga masu fama da cutar sankara 656 sun dace da shawarwarin masana kawai kashi 49 cikin ɗari na lokaci. Likitoci sun tantance hakan Watson bai yi kyau ba tare da tsofaffi marasa lafiyata hanyar rashin ba su wasu daidaitattun magunguna, kuma sun yi babban kuskure na gudanar da aikin sa ido na jiyya ga wasu marasa lafiya da ke fama da ciwon kai.

A ƙarshe, duk da cewa aikinsa na likitan bincike da likita ana ɗaukarsa a matsayin wanda bai yi nasara ba, akwai wuraren da ya tabbatar da cewa yana da matukar amfani. Samfura Watson don Genomics, wanda aka haɓaka tare da haɗin gwiwar Jami'ar North Carolina, Jami'ar Yale, da sauran cibiyoyi, ana amfani da su dakunan gwaje-gwaje na kwayoyin halitta don shirya rahotanni ga masu ilimin oncologists. Watson zazzagewar lissafin fayil maye gurbi a cikin mai haƙuri kuma zai iya samar da rahoto a cikin mintuna wanda ya haɗa da shawarwari ga duk magunguna masu mahimmanci da gwaji na asibiti. Watson yana sarrafa bayanan kwayoyin halitta tare da sauƙin dangisaboda an gabatar da su a cikin fayilolin da aka tsara kuma ba su ƙunshi shubuhohi ba - ko dai akwai maye gurbi ko kuma babu maye gurbi.

Abokan IBM a Jami'ar North Carolina sun buga takarda akan inganci a cikin 2017. Watson ya sami yuwuwar maye gurbi waɗanda binciken ɗan adam bai gano su ba a cikin 32% na su. marasa lafiya sunyi karatu, wanda ya sa su zama 'yan takara masu kyau don sabon magani. Duk da haka, har yanzu babu wata shaida cewa amfani yana haifar da sakamako mafi kyau na magani.

Domestication na sunadaran

Wannan da sauran misalan da yawa suna ba da gudummawa ga haɓakar imani cewa ana magance duk wani lahani na kiwon lafiya, amma muna buƙatar neman wuraren da wannan zai iya taimakawa sosai, saboda mutane ba su da kyau sosai a can. Irin wannan filin shine, alal misali. binciken furotin. A shekarar da ta gabata, bayanai sun bayyana cewa zai iya yin hasashen kamannin sunadaran daidai gwargwadon tsarinsu (2). Wannan aiki ne na al'ada, wanda ya wuce ikon ba kawai mutane ba, har ma da kwamfutoci masu ƙarfi. Idan muka ƙware madaidaicin ƙirar ƙira na karkatar da ƙwayoyin sunadaran gina jiki, za a sami damammaki masu yawa don maganin ƙwayoyin cuta. Masana kimiyya suna fatan cewa tare da taimakon AlphaFold za mu yi nazarin ayyukan dubban mutane, kuma wannan, bi da bi, zai ba mu damar fahimtar abubuwan da ke haifar da cututtuka da yawa.

Hoto 2. Juyawa sunadaran sunadaran ƙira tare da DeepMind's AlphaFold.

Yanzu mun san sunadaran miliyan dari biyu, amma mun fahimci tsari da aikin ƙaramin sashi daga cikinsu. Sunadaran shi ne tushen ginin halittu masu rai. Su ke da alhakin yawancin tafiyar matakai da ke faruwa a cikin sel. Yadda suke aiki da abin da suke yi an ƙaddara ta tsarin su na 50D. Suna ɗaukar sigar da ta dace ba tare da wani umarni ba, waɗanda dokokin kimiyyar lissafi ke jagoranta. Shekaru da yawa, hanyoyin gwaji sun kasance hanya mafi mahimmanci don ƙayyade siffar sunadaran. A cikin XNUMXs, amfani X-ray crystallographic hanyoyin. A cikin shekaru goma da suka gabata, ya zama kayan aikin bincike na zaɓi. crystal microscope. A cikin 80s da 90s, an fara aiki akan amfani da kwamfutoci don tantance siffar sunadaran. Duk da haka, sakamakon har yanzu bai gamsar da masana kimiyya ba. Hanyoyin da ke aiki ga wasu sunadaran ba su yi aiki ga wasu ba.

Tuni a cikin 2018 AlphaFold samu karbuwa daga masana a gina jiki yin samfuri. Duk da haka, a lokacin ya yi amfani da hanyoyi masu kama da sauran shirye-shirye. Masanan kimiyya sun canza dabara kuma suka ƙirƙiri wani, wanda kuma yayi amfani da bayanai game da hani na zahiri da na geometric a cikin nada ƙwayoyin furotin. AlphaFold ya ba da sakamako mara daidaituwa. Wani lokaci ya yi mafi kyau, wani lokacin mafi muni. Amma kusan kashi biyu bisa uku na hasashensa sun zo daidai da sakamakon da aka samu ta hanyoyin gwaji. A farkon shekara ta 2, algorithm ya bayyana tsarin sunadarai da yawa na kwayar cutar SARS-CoV-3. Daga baya, an gano cewa hasashen furotin na Orf2020a ya yi daidai da sakamakon da aka samu ta gwaji.

Ba wai kawai game da nazarin hanyoyin ciki na nadawa sunadaran ba, har ma game da ƙira. Masu bincike daga shirin NIH BRAIN da aka yi amfani da su koyon inji haɓaka furotin wanda zai iya bin matakan serotonin na kwakwalwa a ainihin lokacin. Serotonin wani neurochemical ne wanda ke taka muhimmiyar rawa a yadda kwakwalwa ke sarrafa tunaninmu da ji. Misali, an ƙera magungunan kashe-kashe da yawa don canza siginar siginar serotonin waɗanda ake watsawa tsakanin jijiya. A cikin wata kasida a mujallar Cell, masana kimiyya sun bayyana yadda suke amfani da na gaba hanyoyin injiniyan kwayoyin halitta juya furotin na kwayan cuta zuwa sabon kayan aikin bincike wanda zai iya taimakawa waƙa da watsawar serotonin tare da daidaito mafi girma fiye da hanyoyin yanzu. Gwaje-gwaje na preclinical, galibi a cikin beraye, sun nuna cewa firikwensin zai iya gano canje-canje masu sauƙi a cikin matakan serotonin na kwakwalwa a lokacin barci, tsoro da hulɗar zamantakewa, da gwada tasirin sabbin magungunan psychoactive.

Yaki da cutar ba koyaushe ake samun nasara ba

Bayan haka, wannan ita ce annoba ta farko da muka rubuta game da ita a MT. Duk da haka, alal misali, idan muka yi magana game da ainihin tsarin ci gaban cutar, to a farkon matakin, AI ya zama wani abu na gazawa. Malamai sun koka da cewa Ilimin Artificial ba zai iya yin hasashen daidai girman yaduwar cutar ta coronavirus ba dangane da bayanai daga cututtukan da suka gabata. “Wadannan mafita suna aiki da kyau a wasu fannoni, kamar sanin fuskokin da ke da adadin idanu da kunnuwa. Annobar SARS-CoV-2 Waɗannan al'amuran da ba a san su ba ne a baya da kuma sabbin sauye-sauye da yawa, don haka hankali na wucin gadi bisa bayanan tarihi da aka yi amfani da shi don horar da shi ba ya aiki da kyau. Barkewar cutar ta nuna cewa muna buƙatar neman wasu fasahohi da dabaru, ”in ji Maxim Fedorov daga Skoltech a cikin Afrilu 2020 a cikin wata sanarwa ga kafofin watsa labarai na Rasha.

A tsawon lokaci akwai duk da haka algorithms waɗanda ke da alama suna tabbatar da babban fa'idar AI a cikin yaƙi da COVID-19. Masana kimiyya a Amurka sun haɓaka wani tsari a cikin faɗuwar 2020 don gane halayen tari a cikin mutanen da ke da COVID-19, koda kuwa ba su da wata alama.

Lokacin da alluran rigakafi suka bayyana, an haifi ra'ayin don taimakawa alurar riga kafi. Ta iya, misali taimaka samfurin sufuri da dabaru na rigakafi. Haka kuma wajen tantance yawan jama'a da ya kamata a fara yi wa alurar riga kafi don tunkarar cutar cikin sauri. Hakanan zai taimaka yin hasashen buƙatu da haɓaka lokaci da saurin rigakafin ta hanyar gano matsaloli cikin sauri da cikas a cikin dabaru. Haɗin algorithms tare da saka idanu akai-akai kuma na iya ba da bayanai da sauri game da yiwuwar illa da abubuwan kiwon lafiya.

wadannan tsarin amfani da AI a inganta da inganta kiwon lafiya an riga an san su. An yaba amfaninsu na amfani; misali, tsarin kula da lafiya da Macro-Eyes ya kirkira a Jami’ar Stanford da ke Amurka. Kamar yadda yake a sauran cibiyoyin kiwon lafiya, matsalar ita ce rashin majinyata da ba sa zuwa wurin alƙawura. Macro Idanun ya gina tsarin da zai iya dogara da hasashen abin da marasa lafiya ba za su kasance a wurin ba. A wasu yanayi, yana iya ba da shawarar madadin lokuta da wurare don asibitoci, wanda zai ƙara yuwuwar bayyanar majiyyaci. Daga baya, an yi amfani da irin wannan fasaha a wurare daban-daban daga Arkansas zuwa Najeriya tare da tallafi, musamman, Hukumar Raya Ƙasa ta Amurka i.

A Tanzaniya, Macro-Eyes ya yi aiki a kan wani aikin da aka yi niyya karuwar adadin rigakafin yara. Software ɗin ya bincika adadin alluran rigakafin da ake buƙata don aika zuwa cibiyar rigakafin da aka ba. Ya kuma iya tantance waɗanne iyalai ne za su iya ƙin yi wa ’ya’yansu allurar, amma za a iya shawo kan su da hujjar da ta dace da kuma wurin da za a yi rigakafin a wuri mai kyau. Yin amfani da wannan manhaja, gwamnatin Tanzaniya ta sami damar haɓaka tasirin shirinta na rigakafi da kashi 96%. sannan a rage sharar alluran rigakafin zuwa kashi 2,42 a cikin mutane 100.

A Saliyo, inda bayanan lafiyar mazauna suka ɓace, kamfanin ya yi ƙoƙarin daidaita wannan da bayanai game da ilimi. Sai ya zama cewa adadin malamai da dalibansu kadai ya isa a iya hasashen kashi 70 cikin dari. daidaiton ko cibiyar kula da lafiya ta yankin na da ruwa mai tsafta, wanda tuni ya zama sawun bayanai kan lafiyar mutanen da ke zaune a wurin (3).

3. Hoton Macro-Eyes na shirye-shiryen kiwon lafiya da AI ke tafiyar da shi a Afirka.

Labarin likitan injin ba ya ɓacewa

Duk da kasawa Watson Har yanzu ana ci gaba da haɓaka sabbin hanyoyin gano cutar kuma ana ɗaukar su ƙara haɓakawa. Kwatanta da aka yi a Sweden a cikin Satumba 2020. ana amfani da shi wajen tantance cutar kansar nono ya nuna cewa mafi kyawun su suna aiki kamar yadda likitan rediyo. An gwada algorithms ta amfani da hotuna kusan dubu tara da aka samu yayin tantancewa na yau da kullun. Tsari uku, waɗanda aka tsara azaman AI-1, AI-2 da AI-3, sun sami daidaiton 81,9%, 67%. kuma 67,4%. Don kwatanta, ga masu aikin rediyo waɗanda suka fassara waɗannan hotuna a matsayin na farko, wannan adadi ya kasance 77,4%, kuma a cikin yanayin. masu aikin rediyowanda shi ne na biyu da ya kwatanta shi, ya kai kashi 80,1 bisa dari. Mafi kyawun algorithms kuma ya sami damar gano lamuran da likitocin rediyo suka rasa yayin tantancewa, kuma an gano mata ba su da lafiya cikin ƙasa da shekara guda.

A cewar masu binciken, wadannan sakamakon sun tabbatar da haka Artificial Intelligence algorithms taimakawa wajen gyara cututtukan da ba su dace ba da likitocin rediyo suka yi. Haɗuwa da damar AI-1 tare da matsakaita masanin rediyo ya haɓaka adadin cututtukan nono da aka gano da kashi 8%. Ƙungiyar Royal Institute da ke bayan wannan binciken suna tsammanin ingancin algorithms AI don ci gaba da ingantawa. An buga cikakken bayanin gwajin a cikin JAMA Oncology.

W akan sikelin maki biyar. A halin yanzu, muna ganin babban haɓakar fasaha da kuma isa matakin IV (babban aiki da kai), lokacin da tsarin ke sarrafa bayanan da aka karɓa ta atomatik kuma yana ba ƙwararrun bayanan da aka riga aka bincika. Wannan yana adana lokaci, yana guje wa kuskuren ɗan adam kuma yana ba da ingantaccen kulawar haƙuri. Abin da ya yanke kenan a watannin baya Stan A.I. a fannin likitanci na kusa da shi, prof. Janusz Braziewicz daga Ƙungiyar Polish don Magungunan Nukiliya a cikin wata sanarwa ga Kamfanin Dillancin Labarai na Poland.

4. Na'urar kallon hotunan likita

Algorithms, a cewar masana irin su prof. Brazievichko da ba makawa a cikin wannan masana'antar. Dalili shine saurin haɓakar adadin gwaje-gwajen hoto na bincike. Kawai don lokacin 2000-2010. Yawan gwaje-gwaje na MRI da gwaje-gwaje sun karu sau goma. Abin takaici, adadin ƙwararrun likitocin da za su iya aiwatar da su cikin sauri da dogaro bai ƙaru ba. Haka kuma akwai karancin kwararrun kwararru. Aiwatar da algorithms na tushen AI yana adana lokaci kuma yana ba da damar cikakken daidaitattun hanyoyin, da kuma guje wa kuskuren ɗan adam kuma mafi inganci, keɓaɓɓen jiyya ga marasa lafiya.

Kamar yadda ya kasance, kuma Magungunan likita zai iya amfana ci gaban basirar wucin gadi. Kwararru a wannan fanni za su iya tantance ainihin lokacin mutuwar mamacin ta hanyar nazarin sinadarai na sigar tsutsotsi da sauran halittu da ke cin matattun kyallen jikinsu. Matsala ta taso lokacin da aka haɗa gaurayawan ɓoye daga nau'ikan necrophages daban-daban a cikin bincike. Anan ne koyon injin ke shiga cikin wasa. Masana kimiyya a Jami'ar Albany sun haɓaka hanyar fasaha na wucin gadi wanda ke ba da damar gano nau'in tsutsa da sauri bisa ga "hanyoyin sinadarai". Tawagar ta horar da shirin nasu na kwamfuta ta hanyar amfani da gaurayawan nau'ikan sinadarai daban-daban daga nau'ikan kuda guda shida. Ya zana sa hannun sinadarai na tsutsar tsutsa ta hanyar amfani da ma'auni mai yawa, wanda ke gano sinadarai ta hanyar auna daidai gwargwadon rabo da cajin lantarki na ion.

Don haka, kamar yadda kuke gani, duk da haka AI a matsayin mai binciken bincike ba shi da kyau sosai, yana iya zama da amfani sosai a cikin binciken bincike. Wataƙila muna tsammanin da yawa daga gare ta a wannan matakin, muna tsammanin algorithms waɗanda za su sa likitoci su daina aiki (5). Idan muka duba Ilimin Artificial a zahiri, mai da hankali kan takamaiman fa'idodi masu amfani maimakon na gama-gari, aikinta na likitanci ya sake yin alƙawarin sake.

5. Hangen motar likita